Merge branch 'main' into refactor/table-footer

- Use singular/plural "item"/"items" for count variant units
- Limit pagination sizes to md/lg only
- Remove unused size prop from FooterSelectionModeProps
- Remove unused TableSize type import

backend/alembic/versions/689433b0d8de_add_hook_and_hook_execution_log_tables.py

@@ -1,103 +0,0 @@
-"""add_hook_and_hook_execution_log_tables
-
-Revision ID: 689433b0d8de
-Revises: 93a2e195e25c
-Create Date: 2026-03-13 11:25:06.547474
-
-"""
-
-from alembic import op
-import sqlalchemy as sa
-from sqlalchemy.dialects.postgresql import UUID as PGUUID
-
-
-# revision identifiers, used by Alembic.
-revision = "689433b0d8de"
-down_revision = "93a2e195e25c"
-branch_labels = None
-depends_on = None
-
-
-def upgrade() -> None:
-    op.create_table(
-        "hook",
-        sa.Column("id", sa.Integer(), nullable=False),
-        sa.Column("name", sa.String(), nullable=False),
-        sa.Column(
-            "hook_point",
-            sa.Enum("document_ingestion", "query_processing", native_enum=False),
-            nullable=False,
-        ),
-        sa.Column("endpoint_url", sa.Text(), nullable=True),
-        sa.Column("api_key", sa.LargeBinary(), nullable=True),
-        sa.Column("is_reachable", sa.Boolean(), nullable=True),
-        sa.Column(
-            "fail_strategy",
-            sa.Enum("hard", "soft", native_enum=False),
-            nullable=False,
-        ),
-        sa.Column("timeout_seconds", sa.Float(), nullable=False),
-        sa.Column(
-            "is_active", sa.Boolean(), nullable=False, server_default=sa.text("false")
-        ),
-        sa.Column(
-            "deleted", sa.Boolean(), nullable=False, server_default=sa.text("false")
-        ),
-        sa.Column("creator_id", PGUUID(as_uuid=True), nullable=True),
-        sa.Column(
-            "created_at",
-            sa.DateTime(timezone=True),
-            server_default=sa.text("now()"),
-            nullable=False,
-        ),
-        sa.Column(
-            "updated_at",
-            sa.DateTime(timezone=True),
-            server_default=sa.text("now()"),
-            nullable=False,
-        ),
-        sa.ForeignKeyConstraint(["creator_id"], ["user.id"], ondelete="SET NULL"),
-        sa.PrimaryKeyConstraint("id"),
-    )
-    op.create_index(
-        "ix_hook_one_non_deleted_per_point",
-        "hook",
-        ["hook_point"],
-        unique=True,
-        postgresql_where=sa.text("deleted = false"),
-    )
-
-    op.create_table(
-        "hook_execution_log",
-        sa.Column("id", sa.Integer(), nullable=False),
-        sa.Column("hook_id", sa.Integer(), nullable=False),
-        sa.Column(
-            "is_success",
-            sa.Boolean(),
-            nullable=False,
-        ),
-        sa.Column("error_message", sa.Text(), nullable=True),
-        sa.Column("status_code", sa.Integer(), nullable=True),
-        sa.Column("duration_ms", sa.Integer(), nullable=True),
-        sa.Column(
-            "created_at",
-            sa.DateTime(timezone=True),
-            server_default=sa.text("now()"),
-            nullable=False,
-        ),
-        sa.ForeignKeyConstraint(["hook_id"], ["hook.id"], ondelete="CASCADE"),
-        sa.PrimaryKeyConstraint("id"),
-    )
-    op.create_index("ix_hook_execution_log_hook_id", "hook_execution_log", ["hook_id"])
-    op.create_index(
-        "ix_hook_execution_log_created_at", "hook_execution_log", ["created_at"]
-    )
-
-
-def downgrade() -> None:
-    op.drop_index("ix_hook_execution_log_created_at", table_name="hook_execution_log")
-    op.drop_index("ix_hook_execution_log_hook_id", table_name="hook_execution_log")
-    op.drop_table("hook_execution_log")
-
-    op.drop_index("ix_hook_one_non_deleted_per_point", table_name="hook")
-    op.drop_table("hook")

backend/onyx/chat/llm_loop.py

@@ -36,11 +36,9 @@ from onyx.db.memory import add_memory
 from onyx.db.memory import update_memory_at_index
 from onyx.db.memory import UserMemoryContext
 from onyx.db.models import Persona
-from onyx.llm.constants import LlmProviderNames
 from onyx.llm.interfaces import LLM
 from onyx.llm.interfaces import LLMUserIdentity
 from onyx.llm.interfaces import ToolChoiceOptions
-from onyx.llm.utils import is_true_openai_model
 from onyx.prompts.chat_prompts import IMAGE_GEN_REMINDER
 from onyx.prompts.chat_prompts import OPEN_URL_REMINDER
 from onyx.server.query_and_chat.placement import Placement
@@ -74,70 +72,6 @@ from shared_configs.contextvars import get_current_tenant_id
 logger = setup_logger()
 
 
-class EmptyLLMResponseError(RuntimeError):
-    """Raised when the streamed LLM response completes without a usable answer."""
-
-    def __init__(
-        self,
-        *,
-        provider: str,
-        model: str,
-        tool_choice: ToolChoiceOptions,
-        client_error_msg: str,
-        error_code: str = "EMPTY_LLM_RESPONSE",
-        is_retryable: bool = True,
-    ) -> None:
-        super().__init__(client_error_msg)
-        self.provider = provider
-        self.model = model
-        self.tool_choice = tool_choice
-        self.client_error_msg = client_error_msg
-        self.error_code = error_code
-        self.is_retryable = is_retryable
-
-
-def _build_empty_llm_response_error(
-    llm: LLM,
-    llm_step_result: LlmStepResult,
-    tool_choice: ToolChoiceOptions,
-) -> EmptyLLMResponseError:
-    provider = llm.config.model_provider
-    model = llm.config.model_name
-
-    # OpenAI quota exhaustion has reached us as a streamed "stop" with zero content.
-    # When the stream is completely empty and there is no reasoning/tool output, surface
-    # the likely account-level cause instead of a generic tool-calling error.
-    if (
-        not llm_step_result.reasoning
-        and provider == LlmProviderNames.OPENAI
-        and is_true_openai_model(provider, model)
-    ):
-        return EmptyLLMResponseError(
-            provider=provider,
-            model=model,
-            tool_choice=tool_choice,
-            client_error_msg=(
-                "The selected OpenAI model returned an empty streamed response "
-                "before producing any tokens. This commonly happens when the API "
-                "key or project has no remaining quota or billing is not enabled. "
-                "Verify quota and billing for this key and try again."
-            ),
-            error_code="BUDGET_EXCEEDED",
-            is_retryable=False,
-        )
-
-    return EmptyLLMResponseError(
-        provider=provider,
-        model=model,
-        tool_choice=tool_choice,
-        client_error_msg=(
-            "The selected model returned no final answer before the stream "
-            "completed. No text or tool calls were received from the upstream "
-            "provider."
-        ),
-    )
-
-
 def _looks_like_xml_tool_call_payload(text: str | None) -> bool:
     """Detect XML-style marshaled tool calls emitted as plain text."""
     if not text:
@@ -679,12 +613,7 @@ def run_llm_loop(
             )
             citation_processor.update_citation_mapping(project_citation_mapping)
 
-        llm_step_result = LlmStepResult(
-            reasoning=None,
-            answer=None,
-            tool_calls=None,
-            raw_answer=None,
-        )
+        llm_step_result: LlmStepResult | None = None
 
         # Pass the total budget to construct_message_history, which will handle token allocation
         available_tokens = llm.config.max_input_tokens
@@ -1155,18 +1084,12 @@ def run_llm_loop(
                 # As long as 1 tool with citeable documents is called at any point, we ask the LLM to try to cite
                 should_cite_documents = True
 
-        if not llm_step_result.answer and not llm_step_result.tool_calls:
-            raise _build_empty_llm_response_error(
-                llm=llm,
-                llm_step_result=llm_step_result,
-                tool_choice=tool_choice,
-            )
-
-        if not llm_step_result.answer:
+        if not llm_step_result or not llm_step_result.answer:
             raise RuntimeError(
-                "The LLM did not return a final answer after tool execution. "
-                "Typically this indicates invalid tool-call output, a model/provider mismatch, "
-                "or serving API misconfiguration."
+                "The LLM did not return an answer. "
+                "Typically this is an issue with LLMs that do not support tool calling natively, "
+                "or the model serving API is not configured correctly. "
+                "This may also happen with models that are lower quality outputting invalid tool calls."
             )
 
         emitter.emit(

backend/onyx/chat/llm_step.py

@@ -1013,10 +1013,6 @@ def run_llm_step_pkt_generator(
     accumulated_reasoning = ""
     accumulated_answer = ""
     accumulated_raw_answer = ""
-    stream_chunk_count = 0
-    actionable_chunk_count = 0
-    empty_chunk_count = 0
-    finish_reasons: set[str] = set()
     xml_tool_call_content_filter = _XmlToolCallContentFilter()
 
     processor_state: Any = None
@@ -1149,7 +1145,6 @@ def run_llm_step_pkt_generator(
             user_identity=user_identity,
             timeout_override=timeout_override,
         ):
-            stream_chunk_count += 1
             if packet.usage:
                 usage = packet.usage
                 span_generation.span_data.usage = {
@@ -1159,21 +1154,16 @@ def run_llm_step_pkt_generator(
                     "cache_creation_input_tokens": usage.cache_creation_input_tokens,
                 }
                 # Note: LLM cost tracking is now handled in multi_llm.py
-            finish_reason = packet.choice.finish_reason
-            if finish_reason:
-                finish_reasons.add(str(finish_reason))
             delta = packet.choice.delta
 
             # Weird behavior from some model providers, just log and ignore for now
             if (
-                not delta.content
+                delta.content is None
                 and delta.reasoning_content is None
-                and not delta.tool_calls
+                and delta.tool_calls is None
             ):
-                empty_chunk_count += 1
                 logger.warning(
-                    "LLM packet is empty (no content, reasoning, or tool calls). "
-                    f"finish_reason={finish_reason}. Skipping: {packet}"
+                    f"LLM packet is empty (no contents, reasoning or tool calls). Skipping: {packet}"
                 )
                 continue
 
@@ -1182,8 +1172,6 @@ def run_llm_step_pkt_generator(
                     time.monotonic() - stream_start_time
                 )
                 first_action_recorded = True
-            if _delta_has_action(delta):
-                actionable_chunk_count += 1
 
             if custom_token_processor:
                 # The custom token processor can modify the deltas for specific custom logic
@@ -1319,15 +1307,6 @@ def run_llm_step_pkt_generator(
         else:
             logger.debug("Tool calls: []")
 
-    if actionable_chunk_count == 0:
-        logger.warning(
-            "LLM stream completed with no actionable deltas. "
-            f"chunks={stream_chunk_count}, empty_chunks={empty_chunk_count}, "
-            f"finish_reasons={sorted(finish_reasons)}, "
-            f"provider={llm.config.model_provider}, model={llm.config.model_name}, "
-            f"tool_choice={tool_choice}, tools_sent={len(tool_definitions)}"
-        )
-
     return (
         LlmStepResult(
             reasoning=accumulated_reasoning if accumulated_reasoning else None,

backend/onyx/chat/process_message.py

@@ -29,7 +29,6 @@ from onyx.chat.compression import compress_chat_history
 from onyx.chat.compression import find_summary_for_branch
 from onyx.chat.compression import get_compression_params
 from onyx.chat.emitter import get_default_emitter
-from onyx.chat.llm_loop import EmptyLLMResponseError
 from onyx.chat.llm_loop import run_llm_loop
 from onyx.chat.models import AnswerStream
 from onyx.chat.models import ChatBasicResponse
@@ -926,28 +925,9 @@ def handle_stream_message_objects(
         db_session.rollback()
         return
 
-    except EmptyLLMResponseError as e:
-        stack_trace = traceback.format_exc()
-
-        logger.warning(
-            "LLM returned an empty response "
-            f"(provider={e.provider}, model={e.model}, tool_choice={e.tool_choice})"
-        )
-
-        yield StreamingError(
-            error=e.client_error_msg,
-            stack_trace=stack_trace,
-            error_code=e.error_code,
-            is_retryable=e.is_retryable,
-            details={
-                "model": e.model,
-                "provider": e.provider,
-                "tool_choice": e.tool_choice.value,
-            },
-        )
-        db_session.rollback()
     except Exception as e:
         logger.exception(f"Failed to process chat message due to {e}")
+        error_msg = str(e)
         stack_trace = traceback.format_exc()
 
         if llm:
@@ -1066,46 +1046,10 @@ def llm_loop_completion_handle(
         )
 
 
-_CITATION_LINK_START_PATTERN = re.compile(r"\s*\[\[\d+\]\]\(")
-
-
-def _find_markdown_link_end(text: str, destination_start: int) -> int | None:
-    depth = 0
-    i = destination_start
-
-    while i < len(text):
-        curr = text[i]
-        if curr == "\\":
-            i += 2
-            continue
-
-        if curr == "(":
-            depth += 1
-        elif curr == ")":
-            if depth == 0:
-                return i
-            depth -= 1
-
-        i += 1
-
-    return None
-
-
 def remove_answer_citations(answer: str) -> str:
-    stripped_parts: list[str] = []
-    cursor = 0
+    pattern = r"\s*\[\[\d+\]\]\(http[s]?://[^\s]+\)"
 
-    while match := _CITATION_LINK_START_PATTERN.search(answer, cursor):
-        stripped_parts.append(answer[cursor : match.start()])
-        link_end = _find_markdown_link_end(answer, match.end())
-        if link_end is None:
-            stripped_parts.append(answer[match.start() :])
-            return "".join(stripped_parts)
-
-        cursor = link_end + 1
-
-    stripped_parts.append(answer[cursor:])
-    return "".join(stripped_parts)
+    return re.sub(pattern, "", answer)
 
 
 @log_function_time()
@@ -1143,11 +1087,8 @@ def gather_stream(
         raise ValueError("Message ID is required")
 
     if answer is None:
-        if error_msg is not None:
-            answer = ""
-        else:
-            # This should never be the case as these non-streamed flows do not have a stop-generation signal
-            raise RuntimeError("Answer was not generated")
+        # This should never be the case as these non-streamed flows do not have a stop-generation signal
+        raise RuntimeError("Answer was not generated")
 
     return ChatBasicResponse(
         answer=answer,

backend/onyx/configs/app_configs.py

@@ -318,6 +318,9 @@ VERIFY_CREATE_OPENSEARCH_INDEX_ON_INIT_MT = (
 OPENSEARCH_MIGRATION_GET_VESPA_CHUNKS_PAGE_SIZE = int(
     os.environ.get("OPENSEARCH_MIGRATION_GET_VESPA_CHUNKS_PAGE_SIZE") or 500
 )
+OPENSEARCH_OVERRIDE_DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES = int(
+    os.environ.get("OPENSEARCH_DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES") or 0
+)
 
 VESPA_HOST = os.environ.get("VESPA_HOST") or "localhost"
 # NOTE: this is used if and only if the vespa config server is accessible via a

backend/onyx/db/connector_credential_pair.py

@@ -511,7 +511,7 @@ def add_credential_to_connector(
     user: User,
     connector_id: int,
     credential_id: int,
-    cc_pair_name: str,
+    cc_pair_name: str | None,
     access_type: AccessType,
     groups: list[int] | None,
     auto_sync_options: dict | None = None,

backend/onyx/db/enums.py

@@ -304,13 +304,3 @@ class LLMModelFlowType(str, PyEnum):
     CHAT = "chat"
     VISION = "vision"
     CONTEXTUAL_RAG = "contextual_rag"
-
-
-class HookPoint(str, PyEnum):
-    DOCUMENT_INGESTION = "document_ingestion"
-    QUERY_PROCESSING = "query_processing"
-
-
-class HookFailStrategy(str, PyEnum):
-    HARD = "hard"  # exception propagates, pipeline aborts
-    SOFT = "soft"  # log error, return original input, pipeline continues

backend/onyx/db/hook.py

@@ -1,233 +0,0 @@
-import datetime
-from uuid import UUID
-
-from sqlalchemy import delete
-from sqlalchemy import select
-from sqlalchemy.engine import CursorResult
-from sqlalchemy.exc import IntegrityError
-from sqlalchemy.orm import selectinload
-from sqlalchemy.orm import Session
-
-from onyx.db.constants import UNSET
-from onyx.db.constants import UnsetType
-from onyx.db.enums import HookFailStrategy
-from onyx.db.enums import HookPoint
-from onyx.db.models import Hook
-from onyx.db.models import HookExecutionLog
-from onyx.error_handling.error_codes import OnyxErrorCode
-from onyx.error_handling.exceptions import OnyxError
-
-
-# ── Hook CRUD ────────────────────────────────────────────────────────────
-
-
-def get_hook_by_id(
-    *,
-    db_session: Session,
-    hook_id: int,
-    include_deleted: bool = False,
-    include_creator: bool = False,
-) -> Hook | None:
-    stmt = select(Hook).where(Hook.id == hook_id)
-    if not include_deleted:
-        stmt = stmt.where(Hook.deleted.is_(False))
-    if include_creator:
-        stmt = stmt.options(selectinload(Hook.creator))
-    return db_session.scalar(stmt)
-
-
-def get_non_deleted_hook_by_hook_point(
-    *,
-    db_session: Session,
-    hook_point: HookPoint,
-    include_creator: bool = False,
-) -> Hook | None:
-    stmt = (
-        select(Hook).where(Hook.hook_point == hook_point).where(Hook.deleted.is_(False))
-    )
-    if include_creator:
-        stmt = stmt.options(selectinload(Hook.creator))
-    return db_session.scalar(stmt)
-
-
-def get_hooks(
-    *,
-    db_session: Session,
-    include_deleted: bool = False,
-    include_creator: bool = False,
-) -> list[Hook]:
-    stmt = select(Hook)
-    if not include_deleted:
-        stmt = stmt.where(Hook.deleted.is_(False))
-    if include_creator:
-        stmt = stmt.options(selectinload(Hook.creator))
-    stmt = stmt.order_by(Hook.hook_point, Hook.created_at.desc())
-    return list(db_session.scalars(stmt).all())
-
-
-def create_hook__no_commit(
-    *,
-    db_session: Session,
-    name: str,
-    hook_point: HookPoint,
-    endpoint_url: str | None = None,
-    api_key: str | None = None,
-    fail_strategy: HookFailStrategy,
-    timeout_seconds: float,
-    is_active: bool = False,
-    creator_id: UUID | None = None,
-) -> Hook:
-    """Create a new hook for the given hook point.
-
-    At most one non-deleted hook per hook point is allowed. Raises
-    OnyxError(CONFLICT) if a hook already exists, including under concurrent
-    duplicate creates where the partial unique index fires an IntegrityError.
-    """
-    existing = get_non_deleted_hook_by_hook_point(
-        db_session=db_session, hook_point=hook_point
-    )
-    if existing:
-        raise OnyxError(
-            OnyxErrorCode.CONFLICT,
-            f"A hook for '{hook_point.value}' already exists (id={existing.id}).",
-        )
-
-    hook = Hook(
-        name=name,
-        hook_point=hook_point,
-        endpoint_url=endpoint_url,
-        api_key=api_key,
-        fail_strategy=fail_strategy,
-        timeout_seconds=timeout_seconds,
-        is_active=is_active,
-        creator_id=creator_id,
-    )
-    # Use a savepoint so that a failed insert only rolls back this operation,
-    # not the entire outer transaction.
-    savepoint = db_session.begin_nested()
-    try:
-        db_session.add(hook)
-        savepoint.commit()
-    except IntegrityError as exc:
-        savepoint.rollback()
-        if "ix_hook_one_non_deleted_per_point" in str(exc.orig):
-            raise OnyxError(
-                OnyxErrorCode.CONFLICT,
-                f"A hook for '{hook_point.value}' already exists.",
-            )
-        raise  # re-raise unrelated integrity errors (FK violations, etc.)
-    return hook
-
-
-def update_hook__no_commit(
-    *,
-    db_session: Session,
-    hook_id: int,
-    name: str | None = None,
-    endpoint_url: str | None | UnsetType = UNSET,
-    api_key: str | None | UnsetType = UNSET,
-    fail_strategy: HookFailStrategy | None = None,
-    timeout_seconds: float | None = None,
-    is_active: bool | None = None,
-    is_reachable: bool | None = None,
-    include_creator: bool = False,
-) -> Hook:
-    """Update hook fields.
-
-    Sentinel conventions:
-    - endpoint_url, api_key: pass UNSET to leave unchanged; pass None to clear.
-    - name, fail_strategy, timeout_seconds, is_active, is_reachable: pass None to leave unchanged.
-    """
-    hook = get_hook_by_id(
-        db_session=db_session, hook_id=hook_id, include_creator=include_creator
-    )
-    if hook is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, f"Hook with id {hook_id} not found.")
-
-    if name is not None:
-        hook.name = name
-    if not isinstance(endpoint_url, UnsetType):
-        hook.endpoint_url = endpoint_url
-    if not isinstance(api_key, UnsetType):
-        hook.api_key = api_key  # type: ignore[assignment]  # EncryptedString coerces str → SensitiveValue at the ORM level
-    if fail_strategy is not None:
-        hook.fail_strategy = fail_strategy
-    if timeout_seconds is not None:
-        hook.timeout_seconds = timeout_seconds
-    if is_active is not None:
-        hook.is_active = is_active
-    if is_reachable is not None:
-        hook.is_reachable = is_reachable
-
-    db_session.flush()
-    return hook
-
-
-def delete_hook__no_commit(
-    *,
-    db_session: Session,
-    hook_id: int,
-) -> None:
-    hook = get_hook_by_id(db_session=db_session, hook_id=hook_id)
-    if hook is None:
-        raise OnyxError(OnyxErrorCode.NOT_FOUND, f"Hook with id {hook_id} not found.")
-
-    hook.deleted = True
-    hook.is_active = False
-    db_session.flush()
-
-
-# ── HookExecutionLog CRUD ────────────────────────────────────────────────
-
-
-def create_hook_execution_log__no_commit(
-    *,
-    db_session: Session,
-    hook_id: int,
-    is_success: bool,
-    error_message: str | None = None,
-    status_code: int | None = None,
-    duration_ms: int | None = None,
-) -> HookExecutionLog:
-    log = HookExecutionLog(
-        hook_id=hook_id,
-        is_success=is_success,
-        error_message=error_message,
-        status_code=status_code,
-        duration_ms=duration_ms,
-    )
-    db_session.add(log)
-    db_session.flush()
-    return log
-
-
-def get_hook_execution_logs(
-    *,
-    db_session: Session,
-    hook_id: int,
-    limit: int,
-) -> list[HookExecutionLog]:
-    stmt = (
-        select(HookExecutionLog)
-        .where(HookExecutionLog.hook_id == hook_id)
-        .order_by(HookExecutionLog.created_at.desc())
-        .limit(limit)
-    )
-    return list(db_session.scalars(stmt).all())
-
-
-def cleanup_old_execution_logs__no_commit(
-    *,
-    db_session: Session,
-    max_age_days: int,
-) -> int:
-    """Delete execution logs older than max_age_days. Returns the number of rows deleted."""
-    cutoff = datetime.datetime.now(datetime.timezone.utc) - datetime.timedelta(
-        days=max_age_days
-    )
-    result: CursorResult = db_session.execute(  # type: ignore[assignment]
-        delete(HookExecutionLog)
-        .where(HookExecutionLog.created_at < cutoff)
-        .execution_options(synchronize_session=False)
-    )
-    return result.rowcount

backend/onyx/db/models.py

@@ -64,8 +64,6 @@ from onyx.db.enums import (
     BuildSessionStatus,
     EmbeddingPrecision,
     HierarchyNodeType,
-    HookFailStrategy,
-    HookPoint,
     IndexingMode,
     OpenSearchDocumentMigrationStatus,
     OpenSearchTenantMigrationStatus,
@@ -5180,90 +5178,3 @@ class CacheStore(Base):
     expires_at: Mapped[datetime.datetime | None] = mapped_column(
         DateTime(timezone=True), nullable=True
     )
-
-
-class Hook(Base):
-    """Pairs a HookPoint with a customer-provided API endpoint.
-
-    At most one non-deleted Hook per HookPoint is allowed, enforced by a
-    partial unique index on (hook_point) where deleted=false.
-    """
-
-    __tablename__ = "hook"
-
-    id: Mapped[int] = mapped_column(Integer, primary_key=True)
-    name: Mapped[str] = mapped_column(String, nullable=False)
-    hook_point: Mapped[HookPoint] = mapped_column(
-        Enum(HookPoint, native_enum=False), nullable=False
-    )
-    endpoint_url: Mapped[str | None] = mapped_column(Text, nullable=True)
-    api_key: Mapped[SensitiveValue[str] | None] = mapped_column(
-        EncryptedString(), nullable=True
-    )
-    is_reachable: Mapped[bool | None] = mapped_column(
-        Boolean, nullable=True, default=None
-    )  # null = never validated, true = last check passed, false = last check failed
-    fail_strategy: Mapped[HookFailStrategy] = mapped_column(
-        Enum(HookFailStrategy, native_enum=False),
-        nullable=False,
-        default=HookFailStrategy.HARD,
-    )
-    timeout_seconds: Mapped[float] = mapped_column(Float, nullable=False, default=30.0)
-    is_active: Mapped[bool] = mapped_column(Boolean, nullable=False, default=False)
-    deleted: Mapped[bool] = mapped_column(Boolean, nullable=False, default=False)
-    creator_id: Mapped[UUID | None] = mapped_column(
-        PGUUID(as_uuid=True),
-        ForeignKey("user.id", ondelete="SET NULL"),
-        nullable=True,
-    )
-    created_at: Mapped[datetime.datetime] = mapped_column(
-        DateTime(timezone=True), server_default=func.now(), nullable=False
-    )
-    updated_at: Mapped[datetime.datetime] = mapped_column(
-        DateTime(timezone=True),
-        server_default=func.now(),
-        onupdate=func.now(),
-        nullable=False,
-    )
-
-    creator: Mapped["User | None"] = relationship("User", foreign_keys=[creator_id])
-    execution_logs: Mapped[list["HookExecutionLog"]] = relationship(
-        "HookExecutionLog", back_populates="hook", cascade="all, delete-orphan"
-    )
-
-    __table_args__ = (
-        Index(
-            "ix_hook_one_non_deleted_per_point",
-            "hook_point",
-            unique=True,
-            postgresql_where=(deleted == False),  # noqa: E712
-        ),
-    )
-
-
-class HookExecutionLog(Base):
-    """Records hook executions for health monitoring and debugging.
-
-    Currently only failures are logged; the is_success column exists so
-    success logging can be added later without a schema change.
-    Retention: rows older than 30 days are deleted by a nightly Celery task.
-    """
-
-    __tablename__ = "hook_execution_log"
-
-    id: Mapped[int] = mapped_column(Integer, primary_key=True)
-    hook_id: Mapped[int] = mapped_column(
-        Integer,
-        ForeignKey("hook.id", ondelete="CASCADE"),
-        nullable=False,
-        index=True,
-    )
-    is_success: Mapped[bool] = mapped_column(Boolean, nullable=False, default=False)
-    error_message: Mapped[str | None] = mapped_column(Text, nullable=True)
-    status_code: Mapped[int | None] = mapped_column(Integer, nullable=True)
-    duration_ms: Mapped[int | None] = mapped_column(Integer, nullable=True)
-    created_at: Mapped[datetime.datetime] = mapped_column(
-        DateTime(timezone=True), server_default=func.now(), nullable=False, index=True
-    )
-
-    hook: Mapped["Hook"] = relationship("Hook", back_populates="execution_logs")

backend/onyx/document_index/opensearch/constants.py

@@ -1,23 +1,12 @@
 # Default value for the maximum number of tokens a chunk can hold, if none is
 # specified when creating an index.
-import os
-from enum import Enum
+from onyx.configs.app_configs import (
+    OPENSEARCH_OVERRIDE_DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES,
+)
 
 
 DEFAULT_MAX_CHUNK_SIZE = 512
 
-
-# By default OpenSearch will only return a maximum of this many results in a
-# given search. This value is configurable in the index settings.
-DEFAULT_OPENSEARCH_MAX_RESULT_WINDOW = 10_000
-
-
-# For documents which do not have a value for LAST_UPDATED_FIELD_NAME, we assume
-# that the document was last updated this many days ago for the purpose of time
-# cutoff filtering during retrieval.
-ASSUMED_DOCUMENT_AGE_DAYS = 90
-
-
 # Size of the dynamic list used to consider elements during kNN graph creation.
 # Higher values improve search quality but increase indexing time. Values
 # typically range between 100 - 512.
@@ -37,10 +26,10 @@ M = 32  # Set relatively high for better accuracy.
 # we have a much higher chance of all 10 of the final desired docs showing up
 # and getting scored. In worse situations, the final 10 docs don't even show up
 # as the final 10 (worse than just a miss at the reranking step).
-# Defaults to 100 for now. Initially this defaulted to 750 but we were seeing
-# poor search performance.
-DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES = int(
-    os.environ.get("DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES", 100)
+DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES = (
+    OPENSEARCH_OVERRIDE_DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES
+    if OPENSEARCH_OVERRIDE_DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES > 0
+    else 750
 )
 
 # Number of vectors to examine to decide the top k neighbors for the HNSW
@@ -50,43 +39,23 @@ DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES = int(
 # larger than k, you can provide the size parameter to limit the final number of
 # results to k." from
 # https://docs.opensearch.org/latest/query-dsl/specialized/k-nn/index/#ef_search
-EF_SEARCH = DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES
+EF_SEARCH = DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES
 
+# Since the titles are included in the contents, the embedding matches are
+# heavily downweighted as they act as a boost rather than an independent scoring
+# component.
+SEARCH_TITLE_VECTOR_WEIGHT = 0.1
+SEARCH_CONTENT_VECTOR_WEIGHT = 0.45
+# Single keyword weight for both title and content (merged from former title
+# keyword + content keyword).
+SEARCH_KEYWORD_WEIGHT = 0.45
 
-class HybridSearchSubqueryConfiguration(Enum):
-    TITLE_VECTOR_CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD = 1
-    # Current default.
-    CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD = 2
+# NOTE: It is critical that the order of these weights matches the order of the
+# sub-queries in the hybrid search.
+HYBRID_SEARCH_NORMALIZATION_WEIGHTS = [
+    SEARCH_TITLE_VECTOR_WEIGHT,
+    SEARCH_CONTENT_VECTOR_WEIGHT,
+    SEARCH_KEYWORD_WEIGHT,
+]
 
-
-# Will raise and block application start if HYBRID_SEARCH_SUBQUERY_CONFIGURATION
-# is set but not a valid value. If not set, defaults to
-# CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD.
-HYBRID_SEARCH_SUBQUERY_CONFIGURATION: HybridSearchSubqueryConfiguration = (
-    HybridSearchSubqueryConfiguration(
-        int(os.environ["HYBRID_SEARCH_SUBQUERY_CONFIGURATION"])
-    )
-    if os.environ.get("HYBRID_SEARCH_SUBQUERY_CONFIGURATION", None) is not None
-    else HybridSearchSubqueryConfiguration.CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD
-)
-
-
-class HybridSearchNormalizationPipeline(Enum):
-    # Current default.
-    MIN_MAX = 1
-    # NOTE: Using z-score normalization is better for hybrid search from a
-    # theoretical standpoint. Empirically on a small dataset of up to 10K docs,
-    # it's not very different. Likely more impactful at scale.
-    # https://opensearch.org/blog/introducing-the-z-score-normalization-technique-for-hybrid-search/
-    ZSCORE = 2
-
-
-# Will raise and block application start if HYBRID_SEARCH_NORMALIZATION_PIPELINE
-# is set but not a valid value. If not set, defaults to MIN_MAX.
-HYBRID_SEARCH_NORMALIZATION_PIPELINE: HybridSearchNormalizationPipeline = (
-    HybridSearchNormalizationPipeline(
-        int(os.environ["HYBRID_SEARCH_NORMALIZATION_PIPELINE"])
-    )
-    if os.environ.get("HYBRID_SEARCH_NORMALIZATION_PIPELINE", None) is not None
-    else HybridSearchNormalizationPipeline.MIN_MAX
-)
+assert sum(HYBRID_SEARCH_NORMALIZATION_WEIGHTS) == 1.0

backend/onyx/document_index/opensearch/opensearch_document_index.py

@@ -55,13 +55,16 @@ from onyx.document_index.opensearch.schema import PERSONAS_FIELD_NAME
 from onyx.document_index.opensearch.schema import USER_PROJECTS_FIELD_NAME
 from onyx.document_index.opensearch.search import DocumentQuery
 from onyx.document_index.opensearch.search import (
-    get_min_max_normalization_pipeline_name_and_config,
+    MIN_MAX_NORMALIZATION_PIPELINE_CONFIG,
 )
 from onyx.document_index.opensearch.search import (
-    get_normalization_pipeline_name_and_config,
+    MIN_MAX_NORMALIZATION_PIPELINE_NAME,
 )
 from onyx.document_index.opensearch.search import (
-    get_zscore_normalization_pipeline_name_and_config,
+    ZSCORE_NORMALIZATION_PIPELINE_CONFIG,
+)
+from onyx.document_index.opensearch.search import (
+    ZSCORE_NORMALIZATION_PIPELINE_NAME,
 )
 from onyx.indexing.models import DocMetadataAwareIndexChunk
 from onyx.indexing.models import Document
@@ -100,19 +103,13 @@ def set_cluster_state(client: OpenSearchClient) -> None:
             "is not the first time running Onyx against this instance of OpenSearch, these "
             "settings have likely already been set. Not taking any further action..."
         )
-    min_max_normalization_pipeline_name, min_max_normalization_pipeline_config = (
-        get_min_max_normalization_pipeline_name_and_config()
-    )
-    zscore_normalization_pipeline_name, zscore_normalization_pipeline_config = (
-        get_zscore_normalization_pipeline_name_and_config()
+    client.create_search_pipeline(
+        pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
+        pipeline_body=MIN_MAX_NORMALIZATION_PIPELINE_CONFIG,
     )
     client.create_search_pipeline(
-        pipeline_id=min_max_normalization_pipeline_name,
-        pipeline_body=min_max_normalization_pipeline_config,
-    )
-    client.create_search_pipeline(
-        pipeline_id=zscore_normalization_pipeline_name,
-        pipeline_body=zscore_normalization_pipeline_config,
+        pipeline_id=ZSCORE_NORMALIZATION_PIPELINE_NAME,
+        pipeline_body=ZSCORE_NORMALIZATION_PIPELINE_CONFIG,
     )
 
 
@@ -943,10 +940,14 @@ class OpenSearchDocumentIndex(DocumentIndex):
             index_filters=filters,
             include_hidden=False,
         )
-        normalization_pipeline_name, _ = get_normalization_pipeline_name_and_config()
+        # NOTE: Using z-score normalization here because it's better for hybrid
+        # search from a theoretical standpoint. Empirically on a small dataset
+        # of up to 10K docs, it's not very different. Likely more impactful at
+        # scale.
+        # https://opensearch.org/blog/introducing-the-z-score-normalization-technique-for-hybrid-search/
         search_hits: list[SearchHit[DocumentChunk]] = self._client.search(
             body=query_body,
-            search_pipeline_id=normalization_pipeline_name,
+            search_pipeline_id=ZSCORE_NORMALIZATION_PIPELINE_NAME,
         )
 
         # Good place for a breakpoint to inspect the search hits if you have "explain" enabled.

backend/onyx/document_index/opensearch/search.py

@@ -13,21 +13,10 @@ from onyx.configs.constants import INDEX_SEPARATOR
 from onyx.context.search.models import IndexFilters
 from onyx.context.search.models import Tag
 from onyx.document_index.interfaces_new import TenantState
-from onyx.document_index.opensearch.constants import ASSUMED_DOCUMENT_AGE_DAYS
 from onyx.document_index.opensearch.constants import (
-    DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES,
+    DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES,
 )
-from onyx.document_index.opensearch.constants import (
-    DEFAULT_OPENSEARCH_MAX_RESULT_WINDOW,
-)
-from onyx.document_index.opensearch.constants import (
-    HYBRID_SEARCH_NORMALIZATION_PIPELINE,
-)
-from onyx.document_index.opensearch.constants import (
-    HYBRID_SEARCH_SUBQUERY_CONFIGURATION,
-)
-from onyx.document_index.opensearch.constants import HybridSearchNormalizationPipeline
-from onyx.document_index.opensearch.constants import HybridSearchSubqueryConfiguration
+from onyx.document_index.opensearch.constants import HYBRID_SEARCH_NORMALIZATION_WEIGHTS
 from onyx.document_index.opensearch.schema import ACCESS_CONTROL_LIST_FIELD_NAME
 from onyx.document_index.opensearch.schema import ANCESTOR_HIERARCHY_NODE_IDS_FIELD_NAME
 from onyx.document_index.opensearch.schema import CHUNK_INDEX_FIELD_NAME
@@ -54,113 +43,49 @@ from onyx.document_index.opensearch.schema import USER_PROJECTS_FIELD_NAME
 
 # TODO(andrei): Turn all magic dictionaries to pydantic models.
 
-
-def _get_hybrid_search_normalization_weights() -> list[float]:
-    if (
-        HYBRID_SEARCH_SUBQUERY_CONFIGURATION
-        is HybridSearchSubqueryConfiguration.TITLE_VECTOR_CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD
-    ):
-        # Since the titles are included in the contents, the embedding matches
-        # are heavily downweighted as they act as a boost rather than an
-        # independent scoring component.
-        search_title_vector_weight = 0.1
-        search_content_vector_weight = 0.45
-        # Single keyword weight for both title and content (merged from former
-        # title keyword + content keyword).
-        search_keyword_weight = 0.45
-
-        # NOTE: It is critical that the order of these weights matches the order
-        # of the sub-queries in the hybrid search.
-        hybrid_search_normalization_weights = [
-            search_title_vector_weight,
-            search_content_vector_weight,
-            search_keyword_weight,
-        ]
-    elif (
-        HYBRID_SEARCH_SUBQUERY_CONFIGURATION
-        is HybridSearchSubqueryConfiguration.CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD
-    ):
-        search_content_vector_weight = 0.5
-        # Single keyword weight for both title and content (merged from former
-        # title keyword + content keyword).
-        search_keyword_weight = 0.5
-
-        # NOTE: It is critical that the order of these weights matches the order
-        # of the sub-queries in the hybrid search.
-        hybrid_search_normalization_weights = [
-            search_content_vector_weight,
-            search_keyword_weight,
-        ]
-    else:
-        raise ValueError(
-            f"Bug: Unhandled hybrid search subquery configuration: {HYBRID_SEARCH_SUBQUERY_CONFIGURATION}."
-        )
-
-    assert (
-        sum(hybrid_search_normalization_weights) == 1.0
-    ), "Bug: Hybrid search normalization weights do not sum to 1.0."
-
-    return hybrid_search_normalization_weights
-
-
-def get_min_max_normalization_pipeline_name_and_config() -> tuple[str, dict[str, Any]]:
-    min_max_normalization_pipeline_name = "normalization_pipeline_min_max"
-    min_max_normalization_pipeline_config: dict[str, Any] = {
-        "description": "Normalization for keyword and vector scores using min-max",
-        "phase_results_processors": [
-            {
-                # https://docs.opensearch.org/latest/search-plugins/search-pipelines/normalization-processor/
-                "normalization-processor": {
-                    "normalization": {"technique": "min_max"},
-                    "combination": {
-                        "technique": "arithmetic_mean",
-                        "parameters": {
-                            "weights": _get_hybrid_search_normalization_weights()
-                        },
-                    },
-                }
+MIN_MAX_NORMALIZATION_PIPELINE_NAME = "normalization_pipeline_min_max"
+MIN_MAX_NORMALIZATION_PIPELINE_CONFIG: dict[str, Any] = {
+    "description": "Normalization for keyword and vector scores using min-max",
+    "phase_results_processors": [
+        {
+            # https://docs.opensearch.org/latest/search-plugins/search-pipelines/normalization-processor/
+            "normalization-processor": {
+                "normalization": {"technique": "min_max"},
+                "combination": {
+                    "technique": "arithmetic_mean",
+                    "parameters": {"weights": HYBRID_SEARCH_NORMALIZATION_WEIGHTS},
+                },
             }
-        ],
-    }
-    return min_max_normalization_pipeline_name, min_max_normalization_pipeline_config
+        }
+    ],
+}
 
-
-def get_zscore_normalization_pipeline_name_and_config() -> tuple[str, dict[str, Any]]:
-    zscore_normalization_pipeline_name = "normalization_pipeline_zscore"
-    zscore_normalization_pipeline_config: dict[str, Any] = {
-        "description": "Normalization for keyword and vector scores using z-score",
-        "phase_results_processors": [
-            {
-                # https://docs.opensearch.org/latest/search-plugins/search-pipelines/normalization-processor/
-                "normalization-processor": {
-                    "normalization": {"technique": "z_score"},
-                    "combination": {
-                        "technique": "arithmetic_mean",
-                        "parameters": {
-                            "weights": _get_hybrid_search_normalization_weights()
-                        },
-                    },
-                }
+ZSCORE_NORMALIZATION_PIPELINE_NAME = "normalization_pipeline_zscore"
+ZSCORE_NORMALIZATION_PIPELINE_CONFIG: dict[str, Any] = {
+    "description": "Normalization for keyword and vector scores using z-score",
+    "phase_results_processors": [
+        {
+            # https://docs.opensearch.org/latest/search-plugins/search-pipelines/normalization-processor/
+            "normalization-processor": {
+                "normalization": {"technique": "z_score"},
+                "combination": {
+                    "technique": "arithmetic_mean",
+                    "parameters": {"weights": HYBRID_SEARCH_NORMALIZATION_WEIGHTS},
+                },
             }
-        ],
-    }
-    return zscore_normalization_pipeline_name, zscore_normalization_pipeline_config
+        }
+    ],
+}
 
 
-def get_normalization_pipeline_name_and_config() -> tuple[str, dict[str, Any]]:
-    if (
-        HYBRID_SEARCH_NORMALIZATION_PIPELINE
-        is HybridSearchNormalizationPipeline.MIN_MAX
-    ):
-        return get_min_max_normalization_pipeline_name_and_config()
-    elif (
-        HYBRID_SEARCH_NORMALIZATION_PIPELINE is HybridSearchNormalizationPipeline.ZSCORE
-    ):
-        return get_zscore_normalization_pipeline_name_and_config()
-    else:
-        raise ValueError(
-            f"Bug: Unhandled hybrid search normalization pipeline: {HYBRID_SEARCH_NORMALIZATION_PIPELINE}."
-        )
+# By default OpenSearch will only return a maximum of this many results in a
+# given search. This value is configurable in the index settings.
+DEFAULT_OPENSEARCH_MAX_RESULT_WINDOW = 10_000
+
+# For documents which do not have a value for LAST_UPDATED_FIELD_NAME, we assume
+# that the document was last updated this many days ago for the purpose of time
+# cutoff filtering during retrieval.
+ASSUMED_DOCUMENT_AGE_DAYS = 90
 
 
 class DocumentQuery:
@@ -332,7 +257,7 @@ class DocumentQuery:
 
         # TODO(andrei, yuhong): We can tune this more dynamically based on
         # num_hits.
-        max_results_per_subquery = DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES
+        max_results_per_subquery = DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES
 
         hybrid_search_subqueries = DocumentQuery._get_hybrid_search_subqueries(
             query_text, query_vector, vector_candidates=max_results_per_subquery
@@ -460,7 +385,7 @@ class DocumentQuery:
         # search. This is higher than the number of results because the scoring
         # is hybrid. For a detailed breakdown, see where the default value is
         # set.
-        vector_candidates: int = DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES,
+        vector_candidates: int = DEFAULT_NUM_HYBRID_SEARCH_CANDIDATES,
     ) -> list[dict[str, Any]]:
         """Returns subqueries for hybrid search.
 
@@ -470,18 +395,20 @@ class DocumentQuery:
         The return of this function is not sufficient to be directly supplied to
         the OpenSearch client. See get_hybrid_search_query.
 
+        Matches:
+          - Title vector
+          - Content vector
+          - Keyword (title + content, match and phrase)
+
         Normalization is not performed here.
         The weights of each of these subqueries should be configured in a search
         pipeline.
 
-        The exact subqueries executed depend on the
-        HYBRID_SEARCH_SUBQUERY_CONFIGURATION setting.
-
         NOTE: For OpenSearch, 5 is the maximum number of query clauses allowed
         in a single hybrid query. Source:
         https://docs.opensearch.org/latest/query-dsl/compound/hybrid/
 
-        NOTE: Each query is independent during the search phase; there is no
+        NOTE: Each query is independent during the search phase, there is no
         backfilling of scores for missing query components. What this means is
         that if a document was a good vector match but did not show up for
         keyword, it gets a score of 0 for the keyword component of the hybrid
@@ -510,115 +437,74 @@ class DocumentQuery:
                 similarity search.
         """
         # Build sub-queries for hybrid search. Order must match normalization
-        # pipeline weights.
-        if (
-            HYBRID_SEARCH_SUBQUERY_CONFIGURATION
-            is HybridSearchSubqueryConfiguration.TITLE_VECTOR_CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD
-        ):
-            return [
-                DocumentQuery._get_title_vector_similarity_search_query(
-                    query_vector, vector_candidates
-                ),
-                DocumentQuery._get_content_vector_similarity_search_query(
-                    query_vector, vector_candidates
-                ),
-                DocumentQuery._get_title_content_combined_keyword_search_query(
-                    query_text
-                ),
-            ]
-        elif (
-            HYBRID_SEARCH_SUBQUERY_CONFIGURATION
-            is HybridSearchSubqueryConfiguration.CONTENT_VECTOR_TITLE_CONTENT_COMBINED_KEYWORD
-        ):
-            return [
-                DocumentQuery._get_content_vector_similarity_search_query(
-                    query_vector, vector_candidates
-                ),
-                DocumentQuery._get_title_content_combined_keyword_search_query(
-                    query_text
-                ),
-            ]
-        else:
-            raise ValueError(
-                f"Bug: Unhandled hybrid search subquery configuration: {HYBRID_SEARCH_SUBQUERY_CONFIGURATION}"
-            )
-
-    @staticmethod
-    def _get_title_vector_similarity_search_query(
-        query_vector: list[float],
-        vector_candidates: int = DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES,
-    ) -> dict[str, Any]:
-        return {
-            "knn": {
-                TITLE_VECTOR_FIELD_NAME: {
-                    "vector": query_vector,
-                    "k": vector_candidates,
+        # pipeline weights: title vector, content vector, keyword (title + content).
+        hybrid_search_queries: list[dict[str, Any]] = [
+            # 1. Title vector search
+            {
+                "knn": {
+                    TITLE_VECTOR_FIELD_NAME: {
+                        "vector": query_vector,
+                        "k": vector_candidates,
+                    }
                 }
-            }
-        }
-
-    @staticmethod
-    def _get_content_vector_similarity_search_query(
-        query_vector: list[float],
-        vector_candidates: int = DEFAULT_NUM_HYBRID_SUBQUERY_CANDIDATES,
-    ) -> dict[str, Any]:
-        return {
-            "knn": {
-                CONTENT_VECTOR_FIELD_NAME: {
-                    "vector": query_vector,
-                    "k": vector_candidates,
+            },
+            # 2. Content vector search
+            {
+                "knn": {
+                    CONTENT_VECTOR_FIELD_NAME: {
+                        "vector": query_vector,
+                        "k": vector_candidates,
+                    }
                 }
-            }
-        }
+            },
+            # 3. Keyword (title + content) match and phrase search.
+            {
+                "bool": {
+                    "should": [
+                        {
+                            "match": {
+                                TITLE_FIELD_NAME: {
+                                    "query": query_text,
+                                    "operator": "or",
+                                    # The title fields are strongly discounted as they are included in the content.
+                                    # It just acts as a minor boost
+                                    "boost": 0.1,
+                                }
+                            }
+                        },
+                        {
+                            "match_phrase": {
+                                TITLE_FIELD_NAME: {
+                                    "query": query_text,
+                                    "slop": 1,
+                                    "boost": 0.2,
+                                }
+                            }
+                        },
+                        {
+                            "match": {
+                                CONTENT_FIELD_NAME: {
+                                    "query": query_text,
+                                    "operator": "or",
+                                    "boost": 1.0,
+                                }
+                            }
+                        },
+                        {
+                            "match_phrase": {
+                                CONTENT_FIELD_NAME: {
+                                    "query": query_text,
+                                    "slop": 1,
+                                    "boost": 1.5,
+                                }
+                            }
+                        },
+                    ]
+                }
+            },
+        ]
 
-    @staticmethod
-    def _get_title_content_combined_keyword_search_query(
-        query_text: str,
-    ) -> dict[str, Any]:
-        return {
-            "bool": {
-                "should": [
-                    {
-                        "match": {
-                            TITLE_FIELD_NAME: {
-                                "query": query_text,
-                                "operator": "or",
-                                # The title fields are strongly discounted as they are included in the content.
-                                # It just acts as a minor boost
-                                "boost": 0.1,
-                            }
-                        }
-                    },
-                    {
-                        "match_phrase": {
-                            TITLE_FIELD_NAME: {
-                                "query": query_text,
-                                "slop": 1,
-                                "boost": 0.2,
-                            }
-                        }
-                    },
-                    {
-                        "match": {
-                            CONTENT_FIELD_NAME: {
-                                "query": query_text,
-                                "operator": "or",
-                                "boost": 1.0,
-                            }
-                        }
-                    },
-                    {
-                        "match_phrase": {
-                            CONTENT_FIELD_NAME: {
-                                "query": query_text,
-                                "slop": 1,
-                                "boost": 1.5,
-                            }
-                        }
-                    },
-                ]
-            }
-        }
+        return hybrid_search_queries
 
     @staticmethod
     def _get_search_filters(

backend/onyx/document_index/vespa/chunk_retrieval.py

@@ -501,31 +501,20 @@ def query_vespa(
             response = http_client.post(SEARCH_ENDPOINT, json=params)
             response.raise_for_status()
     except httpx.HTTPError as e:
-        response_text = (
-            e.response.text if isinstance(e, httpx.HTTPStatusError) else None
-        )
-        status_code = (
-            e.response.status_code if isinstance(e, httpx.HTTPStatusError) else None
-        )
-        yql_value = params.get("yql", "")
-        yql_length = len(str(yql_value))
-
-        # Log each detail on its own line so log collectors capture them
-        # as separate entries rather than truncating a single multiline msg
+        error_base = "Failed to query Vespa"
         logger.error(
-            f"Failed to query Vespa | "
-            f"status={status_code} | "
-            f"yql_length={yql_length} | "
-            f"exception={str(e)}"
+            f"{error_base}:\n"
+            f"Request URL: {e.request.url}\n"
+            f"Request Headers: {e.request.headers}\n"
+            f"Request Payload: {params}\n"
+            f"Exception: {str(e)}"
+            + (
+                f"\nResponse: {e.response.text}"
+                if isinstance(e, httpx.HTTPStatusError)
+                else ""
+            )
         )
-        if response_text:
-            logger.error(f"Vespa error response: {response_text[:1000]}")
-        logger.error(f"Vespa request URL: {e.request.url}")
-
-        # Re-raise with diagnostics so callers see what actually went wrong
-        raise httpx.HTTPError(
-            f"Failed to query Vespa (status={status_code}, " f"yql_length={yql_length})"
-        ) from e
+        raise httpx.HTTPError(error_base) from e
 
     response_json: dict[str, Any] = response.json()
 

backend/onyx/document_index/vespa/shared_utils/vespa_request_builders.py

@@ -43,22 +43,6 @@ def build_vespa_filters(
             return ""
         return f"({' or '.join(eq_elems)})"
 
-    def _build_weighted_set_filter(key: str, vals: list[str] | None) -> str:
-        """Build a Vespa weightedSet filter for large value lists.
-
-        Uses Vespa's native weightedSet() operator instead of OR-chained
-        'contains' clauses.  This is critical for fields like
-        access_control_list where a single user may have tens of thousands
-        of ACL entries — OR clauses at that scale cause Vespa to reject
-        the query with HTTP 400."""
-        if not key or not vals:
-            return ""
-        filtered = [val for val in vals if val]
-        if not filtered:
-            return ""
-        items = ", ".join(f'"{val}":1' for val in filtered)
-        return f"weightedSet({key}, {{{items}}})"
-
     def _build_int_or_filters(key: str, vals: list[int] | None) -> str:
         """For an integer field filter.
         Returns a bare clause or ""."""
@@ -173,16 +157,11 @@ def build_vespa_filters(
     if filters.tenant_id and MULTI_TENANT:
         filter_parts.append(build_tenant_id_filter(filters.tenant_id))
 
-    # ACL filters — use weightedSet for efficient matching against the
-    # access_control_list weightedset<string> field.  OR-chaining thousands
-    # of 'contains' clauses causes Vespa to reject the query (HTTP 400)
-    # for users with large numbers of external permission groups.
+    # ACL filters
     if filters.access_control_list is not None:
         _append(
             filter_parts,
-            _build_weighted_set_filter(
-                ACCESS_CONTROL_LIST, filters.access_control_list
-            ),
+            _build_or_filters(ACCESS_CONTROL_LIST, filters.access_control_list),
         )
 
     # Source type filters

backend/onyx/llm/utils.py

@@ -219,26 +219,13 @@ def litellm_exception_to_error_msg(
             "ratelimiterror"
         ):
             upstream_detail = upstream_detail.split(":", 1)[1].strip()
-        upstream_detail_lower = upstream_detail.lower()
-        if (
-            "insufficient_quota" in upstream_detail_lower
-            or "exceeded your current quota" in upstream_detail_lower
-        ):
-            error_msg = (
-                f"{provider_name} quota exceeded: {upstream_detail}"
-                if upstream_detail
-                else f"{provider_name} quota exceeded: Verify billing and quota for this API key."
-            )
-            error_code = "BUDGET_EXCEEDED"
-            is_retryable = False
-        else:
-            error_msg = (
-                f"{provider_name} rate limit: {upstream_detail}"
-                if upstream_detail
-                else f"{provider_name} rate limit exceeded: Please slow down your requests and try again later."
-            )
-            error_code = "RATE_LIMIT"
-            is_retryable = True
+        error_msg = (
+            f"{provider_name} rate limit: {upstream_detail}"
+            if upstream_detail
+            else f"{provider_name} rate limit exceeded: Please slow down your requests and try again later."
+        )
+        error_code = "RATE_LIMIT"
+        is_retryable = True
     elif isinstance(core_exception, ServiceUnavailableError):
         provider_name = (
             llm.config.model_provider

backend/onyx/server/documents/models.py

@@ -408,7 +408,7 @@ class FailedConnectorIndexingStatus(BaseModel):
     """Simplified version of ConnectorIndexingStatus for failed indexing attempts"""
 
     cc_pair_id: int
-    name: str
+    name: str | None
     error_msg: str | None
     is_deletable: bool
     connector_id: int
@@ -422,7 +422,7 @@ class ConnectorStatus(BaseModel):
     """
 
     cc_pair_id: int
-    name: str
+    name: str | None
     connector: ConnectorSnapshot
     credential: CredentialSnapshot
     access_type: AccessType
@@ -453,7 +453,7 @@ class DocsCountOperator(str, Enum):
 
 class ConnectorIndexingStatusLite(BaseModel):
     cc_pair_id: int
-    name: str
+    name: str | None
     source: DocumentSource
     access_type: AccessType
     cc_pair_status: ConnectorCredentialPairStatus
@@ -488,7 +488,7 @@ class ConnectorCredentialPairIdentifier(BaseModel):
 
 
 class ConnectorCredentialPairMetadata(BaseModel):
-    name: str
+    name: str | None = None
     access_type: AccessType
     auto_sync_options: dict[str, Any] | None = None
     groups: list[int] = Field(default_factory=list)
@@ -501,7 +501,7 @@ class CCStatusUpdateRequest(BaseModel):
 
 class ConnectorCredentialPairDescriptor(BaseModel):
     id: int
-    name: str
+    name: str | None = None
     connector: ConnectorSnapshot
     credential: CredentialSnapshot
     access_type: AccessType
@@ -511,7 +511,7 @@ class CCPairSummary(BaseModel):
     """Simplified connector-credential pair information with just essential data"""
 
     id: int
-    name: str
+    name: str | None
     source: DocumentSource
     access_type: AccessType
 

backend/onyx/server/features/build/sandbox/kubernetes/docker/templates/outputs/web/package-lock.json

@@ -15,7 +15,7 @@
         "date-fns": "^4.1.0",
         "embla-carousel-react": "^8.6.0",
         "lucide-react": "^0.562.0",
-        "next": "16.1.7",
+        "next": "16.1.5",
         "next-themes": "^0.4.6",
         "radix-ui": "^1.4.3",
         "react": "19.2.3",
@@ -1711,9 +1711,9 @@
       }
     },
     "node_modules/@next/env": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/env/-/env-16.1.7.tgz",
-      "integrity": "sha512-rJJbIdJB/RQr2F1nylZr/PJzamvNNhfr3brdKP6s/GW850jbtR70QlSfFselvIBbcPUOlQwBakexjFzqLzF6pg==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/env/-/env-16.1.5.tgz",
+      "integrity": "sha512-CRSCPJiSZoi4Pn69RYBDI9R7YK2g59vLexPQFXY0eyw+ILevIenCywzg+DqmlBik9zszEnw2HLFOUlLAcJbL7g==",
       "license": "MIT"
     },
     "node_modules/@next/eslint-plugin-next": {
@@ -1727,9 +1727,9 @@
       }
     },
     "node_modules/@next/swc-darwin-arm64": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-darwin-arm64/-/swc-darwin-arm64-16.1.7.tgz",
-      "integrity": "sha512-b2wWIE8sABdyafc4IM8r5Y/dS6kD80JRtOGrUiKTsACFQfWWgUQ2NwoUX1yjFMXVsAwcQeNpnucF2ZrujsBBPg==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-darwin-arm64/-/swc-darwin-arm64-16.1.5.tgz",
+      "integrity": "sha512-eK7Wdm3Hjy/SCL7TevlH0C9chrpeOYWx2iR7guJDaz4zEQKWcS1IMVfMb9UKBFMg1XgzcPTYPIp1Vcpukkjg6Q==",
       "cpu": [
         "arm64"
       ],
@@ -1743,9 +1743,9 @@
       }
     },
     "node_modules/@next/swc-darwin-x64": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-darwin-x64/-/swc-darwin-x64-16.1.7.tgz",
-      "integrity": "sha512-zcnVaaZulS1WL0Ss38R5Q6D2gz7MtBu8GZLPfK+73D/hp4GFMrC2sudLky1QibfV7h6RJBJs/gOFvYP0X7UVlQ==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-darwin-x64/-/swc-darwin-x64-16.1.5.tgz",
+      "integrity": "sha512-foQscSHD1dCuxBmGkbIr6ScAUF6pRoDZP6czajyvmXPAOFNnQUJu2Os1SGELODjKp/ULa4fulnBWoHV3XdPLfA==",
       "cpu": [
         "x64"
       ],
@@ -1759,9 +1759,9 @@
       }
     },
     "node_modules/@next/swc-linux-arm64-gnu": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-gnu/-/swc-linux-arm64-gnu-16.1.7.tgz",
-      "integrity": "sha512-2ant89Lux/Q3VyC8vNVg7uBaFVP9SwoK2jJOOR0L8TQnX8CAYnh4uctAScy2Hwj2dgjVHqHLORQZJ2wH6VxhSQ==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-gnu/-/swc-linux-arm64-gnu-16.1.5.tgz",
+      "integrity": "sha512-qNIb42o3C02ccIeSeKjacF3HXotGsxh/FMk/rSRmCzOVMtoWH88odn2uZqF8RLsSUWHcAqTgYmPD3pZ03L9ZAA==",
       "cpu": [
         "arm64"
       ],
@@ -1775,9 +1775,9 @@
       }
     },
     "node_modules/@next/swc-linux-arm64-musl": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-musl/-/swc-linux-arm64-musl-16.1.7.tgz",
-      "integrity": "sha512-uufcze7LYv0FQg9GnNeZ3/whYfo+1Q3HnQpm16o6Uyi0OVzLlk2ZWoY7j07KADZFY8qwDbsmFnMQP3p3+Ftprw==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-arm64-musl/-/swc-linux-arm64-musl-16.1.5.tgz",
+      "integrity": "sha512-U+kBxGUY1xMAzDTXmuVMfhaWUZQAwzRaHJ/I6ihtR5SbTVUEaDRiEU9YMjy1obBWpdOBuk1bcm+tsmifYSygfw==",
       "cpu": [
         "arm64"
       ],
@@ -1791,9 +1791,9 @@
       }
     },
     "node_modules/@next/swc-linux-x64-gnu": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-gnu/-/swc-linux-x64-gnu-16.1.7.tgz",
-      "integrity": "sha512-KWVf2gxYvHtvuT+c4MBOGxuse5TD7DsMFYSxVxRBnOzok/xryNeQSjXgxSv9QpIVlaGzEn/pIuI6Koosx8CGWA==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-gnu/-/swc-linux-x64-gnu-16.1.5.tgz",
+      "integrity": "sha512-gq2UtoCpN7Ke/7tKaU7i/1L7eFLfhMbXjNghSv0MVGF1dmuoaPeEVDvkDuO/9LVa44h5gqpWeJ4mRRznjDv7LA==",
       "cpu": [
         "x64"
       ],
@@ -1807,9 +1807,9 @@
       }
     },
     "node_modules/@next/swc-linux-x64-musl": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-musl/-/swc-linux-x64-musl-16.1.7.tgz",
-      "integrity": "sha512-HguhaGwsGr1YAGs68uRKc4aGWxLET+NevJskOcCAwXbwj0fYX0RgZW2gsOCzr9S11CSQPIkxmoSbuVaBp4Z3dA==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-linux-x64-musl/-/swc-linux-x64-musl-16.1.5.tgz",
+      "integrity": "sha512-bQWSE729PbXT6mMklWLf8dotislPle2L70E9q6iwETYEOt092GDn0c+TTNj26AjmeceSsC4ndyGsK5nKqHYXjQ==",
       "cpu": [
         "x64"
       ],
@@ -1823,9 +1823,9 @@
       }
     },
     "node_modules/@next/swc-win32-arm64-msvc": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-win32-arm64-msvc/-/swc-win32-arm64-msvc-16.1.7.tgz",
-      "integrity": "sha512-S0n3KrDJokKTeFyM/vGGGR8+pCmXYrjNTk2ZozOL1C/JFdfUIL9O1ATaJOl5r2POe56iRChbsszrjMAdWSv7kQ==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-win32-arm64-msvc/-/swc-win32-arm64-msvc-16.1.5.tgz",
+      "integrity": "sha512-LZli0anutkIllMtTAWZlDqdfvjWX/ch8AFK5WgkNTvaqwlouiD1oHM+WW8RXMiL0+vAkAJyAGEzPPjO+hnrSNQ==",
       "cpu": [
         "arm64"
       ],
@@ -1839,9 +1839,9 @@
       }
     },
     "node_modules/@next/swc-win32-x64-msvc": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/@next/swc-win32-x64-msvc/-/swc-win32-x64-msvc-16.1.7.tgz",
-      "integrity": "sha512-mwgtg8CNZGYm06LeEd+bNnOUfwOyNem/rOiP14Lsz+AnUY92Zq/LXwtebtUiaeVkhbroRCQ0c8GlR4UT1U+0yg==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/@next/swc-win32-x64-msvc/-/swc-win32-x64-msvc-16.1.5.tgz",
+      "integrity": "sha512-7is37HJTNQGhjPpQbkKjKEboHYQnCgpVt/4rBrrln0D9nderNxZ8ZWs8w1fAtzUx7wEyYjQ+/13myFgFj6K2Ng==",
       "cpu": [
         "x64"
       ],
@@ -4971,15 +4971,12 @@
       "license": "MIT"
     },
     "node_modules/baseline-browser-mapping": {
-      "version": "2.10.8",
-      "resolved": "https://registry.npmjs.org/baseline-browser-mapping/-/baseline-browser-mapping-2.10.8.tgz",
-      "integrity": "sha512-PCLz/LXGBsNTErbtB6i5u4eLpHeMfi93aUv5duMmj6caNu6IphS4q6UevDnL36sZQv9lrP11dbPKGMaXPwMKfQ==",
+      "version": "2.9.17",
+      "resolved": "https://registry.npmjs.org/baseline-browser-mapping/-/baseline-browser-mapping-2.9.17.tgz",
+      "integrity": "sha512-agD0MgJFUP/4nvjqzIB29zRPUuCF7Ge6mEv9s8dHrtYD7QWXRcx75rOADE/d5ah1NI+0vkDl0yorDd5U852IQQ==",
       "license": "Apache-2.0",
       "bin": {
-        "baseline-browser-mapping": "dist/cli.cjs"
-      },
-      "engines": {
-        "node": ">=6.0.0"
+        "baseline-browser-mapping": "dist/cli.js"
       }
     },
     "node_modules/body-parser": {
@@ -8978,14 +8975,14 @@
       }
     },
     "node_modules/next": {
-      "version": "16.1.7",
-      "resolved": "https://registry.npmjs.org/next/-/next-16.1.7.tgz",
-      "integrity": "sha512-WM0L7WrSvKwoLegLYr6V+mz+RIofqQgVAfHhMp9a88ms0cFX8iX9ew+snpWlSBwpkURJOUdvCEt3uLl3NNzvWg==",
+      "version": "16.1.5",
+      "resolved": "https://registry.npmjs.org/next/-/next-16.1.5.tgz",
+      "integrity": "sha512-f+wE+NSbiQgh3DSAlTaw2FwY5yGdVViAtp8TotNQj4kk4Q8Bh1sC/aL9aH+Rg1YAVn18OYXsRDT7U/079jgP7w==",
       "license": "MIT",
       "dependencies": {
-        "@next/env": "16.1.7",
+        "@next/env": "16.1.5",
         "@swc/helpers": "0.5.15",
-        "baseline-browser-mapping": "^2.9.19",
+        "baseline-browser-mapping": "^2.8.3",
         "caniuse-lite": "^1.0.30001579",
         "postcss": "8.4.31",
         "styled-jsx": "5.1.6"
@@ -8997,14 +8994,14 @@
         "node": ">=20.9.0"
       },
       "optionalDependencies": {
-        "@next/swc-darwin-arm64": "16.1.7",
-        "@next/swc-darwin-x64": "16.1.7",
-        "@next/swc-linux-arm64-gnu": "16.1.7",
-        "@next/swc-linux-arm64-musl": "16.1.7",
-        "@next/swc-linux-x64-gnu": "16.1.7",
-        "@next/swc-linux-x64-musl": "16.1.7",
-        "@next/swc-win32-arm64-msvc": "16.1.7",
-        "@next/swc-win32-x64-msvc": "16.1.7",
+        "@next/swc-darwin-arm64": "16.1.5",
+        "@next/swc-darwin-x64": "16.1.5",
+        "@next/swc-linux-arm64-gnu": "16.1.5",
+        "@next/swc-linux-arm64-musl": "16.1.5",
+        "@next/swc-linux-x64-gnu": "16.1.5",
+        "@next/swc-linux-x64-musl": "16.1.5",
+        "@next/swc-win32-arm64-msvc": "16.1.5",
+        "@next/swc-win32-x64-msvc": "16.1.5",
         "sharp": "^0.34.4"
       },
       "peerDependencies": {

backend/onyx/server/features/build/sandbox/kubernetes/docker/templates/outputs/web/package.json

@@ -16,7 +16,7 @@
     "date-fns": "^4.1.0",
     "embla-carousel-react": "^8.6.0",
     "lucide-react": "^0.562.0",
-    "next": "16.1.7",
+    "next": "16.1.5",
     "next-themes": "^0.4.6",
     "radix-ui": "^1.4.3",
     "react": "19.2.3",

backend/onyx/server/query_and_chat/streaming_utils.py

@@ -0,0 +1,239 @@
+from __future__ import annotations
+
+import re
+
+from onyx.context.search.models import SavedSearchDoc
+from onyx.context.search.models import SearchDoc
+from onyx.server.query_and_chat.placement import Placement
+from onyx.server.query_and_chat.streaming_models import AgentResponseDelta
+from onyx.server.query_and_chat.streaming_models import AgentResponseStart
+from onyx.server.query_and_chat.streaming_models import CitationInfo
+from onyx.server.query_and_chat.streaming_models import GeneratedImage
+from onyx.server.query_and_chat.streaming_models import ImageGenerationFinal
+from onyx.server.query_and_chat.streaming_models import ImageGenerationToolStart
+from onyx.server.query_and_chat.streaming_models import OpenUrlDocuments
+from onyx.server.query_and_chat.streaming_models import OpenUrlStart
+from onyx.server.query_and_chat.streaming_models import OpenUrlUrls
+from onyx.server.query_and_chat.streaming_models import Packet
+from onyx.server.query_and_chat.streaming_models import ReasoningDelta
+from onyx.server.query_and_chat.streaming_models import ReasoningStart
+from onyx.server.query_and_chat.streaming_models import SearchToolDocumentsDelta
+from onyx.server.query_and_chat.streaming_models import SearchToolQueriesDelta
+from onyx.server.query_and_chat.streaming_models import SearchToolStart
+from onyx.server.query_and_chat.streaming_models import SectionEnd
+
+
+_CANNOT_SHOW_STEP_RESULTS_STR = "[Cannot display step results]"
+
+
+def _adjust_message_text_for_agent_search_results(
+    adjusted_message_text: str,
+    final_documents: list[SavedSearchDoc],  # noqa: ARG001
+) -> str:
+    # Remove all [Q<integer>] patterns (sub-question citations)
+    return re.sub(r"\[Q\d+\]", "", adjusted_message_text)
+
+
+def _replace_d_citations_with_links(
+    message_text: str, final_documents: list[SavedSearchDoc]
+) -> str:
+    def replace_citation(match: re.Match[str]) -> str:
+        d_number = match.group(1)
+        try:
+            doc_index = int(d_number) - 1
+            if 0 <= doc_index < len(final_documents):
+                doc = final_documents[doc_index]
+                link = doc.link if doc.link else ""
+                return f"[[{d_number}]]({link})"
+            return match.group(0)
+        except (ValueError, IndexError):
+            return match.group(0)
+
+    return re.sub(r"\[D(\d+)\]", replace_citation, message_text)
+
+
+def create_message_packets(
+    message_text: str,
+    final_documents: list[SavedSearchDoc] | None,
+    turn_index: int,
+    is_legacy_agentic: bool = False,
+) -> list[Packet]:
+    packets: list[Packet] = []
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=AgentResponseStart(
+                final_documents=SearchDoc.from_saved_search_docs(final_documents or []),
+            ),
+        )
+    )
+
+    adjusted_message_text = message_text
+    if is_legacy_agentic:
+        if final_documents is not None:
+            adjusted_message_text = _adjust_message_text_for_agent_search_results(
+                message_text, final_documents
+            )
+            adjusted_message_text = _replace_d_citations_with_links(
+                adjusted_message_text, final_documents
+            )
+        else:
+            adjusted_message_text = re.sub(r"\[Q\d+\]", "", message_text)
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=AgentResponseDelta(
+                content=adjusted_message_text,
+            ),
+        ),
+    )
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=SectionEnd(),
+        )
+    )
+
+    return packets
+
+
+def create_citation_packets(
+    citation_info_list: list[CitationInfo], turn_index: int
+) -> list[Packet]:
+    packets: list[Packet] = []
+
+    # Emit each citation as a separate CitationInfo packet
+    for citation_info in citation_info_list:
+        packets.append(
+            Packet(
+                placement=Placement(turn_index=turn_index),
+                obj=citation_info,
+            )
+        )
+
+    packets.append(Packet(placement=Placement(turn_index=turn_index), obj=SectionEnd()))
+
+    return packets
+
+
+def create_reasoning_packets(reasoning_text: str, turn_index: int) -> list[Packet]:
+    packets: list[Packet] = []
+
+    packets.append(
+        Packet(placement=Placement(turn_index=turn_index), obj=ReasoningStart())
+    )
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=ReasoningDelta(
+                reasoning=reasoning_text,
+            ),
+        ),
+    )
+
+    packets.append(Packet(placement=Placement(turn_index=turn_index), obj=SectionEnd()))
+
+    return packets
+
+
+def create_image_generation_packets(
+    images: list[GeneratedImage], turn_index: int
+) -> list[Packet]:
+    packets: list[Packet] = []
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=ImageGenerationToolStart(),
+        )
+    )
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=ImageGenerationFinal(images=images),
+        ),
+    )
+
+    packets.append(Packet(placement=Placement(turn_index=turn_index), obj=SectionEnd()))
+
+    return packets
+
+
+def create_fetch_packets(
+    fetch_docs: list[SavedSearchDoc],
+    urls: list[str],
+    turn_index: int,
+) -> list[Packet]:
+    packets: list[Packet] = []
+    # Emit start packet
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=OpenUrlStart(),
+        )
+    )
+    # Emit URLs packet
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=OpenUrlUrls(urls=urls),
+        )
+    )
+    # Emit documents packet
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=OpenUrlDocuments(
+                documents=SearchDoc.from_saved_search_docs(fetch_docs)
+            ),
+        )
+    )
+    packets.append(Packet(placement=Placement(turn_index=turn_index), obj=SectionEnd()))
+    return packets
+
+
+def create_search_packets(
+    search_queries: list[str],
+    saved_search_docs: list[SavedSearchDoc],
+    is_internet_search: bool,
+    turn_index: int,
+) -> list[Packet]:
+    packets: list[Packet] = []
+
+    packets.append(
+        Packet(
+            placement=Placement(turn_index=turn_index),
+            obj=SearchToolStart(
+                is_internet_search=is_internet_search,
+            ),
+        )
+    )
+
+    # Emit queries if present
+    if search_queries:
+        packets.append(
+            Packet(
+                placement=Placement(turn_index=turn_index),
+                obj=SearchToolQueriesDelta(queries=search_queries),
+            ),
+        )
+
+    # Emit documents if present
+    if saved_search_docs:
+        packets.append(
+            Packet(
+                placement=Placement(turn_index=turn_index),
+                obj=SearchToolDocumentsDelta(
+                    documents=SearchDoc.from_saved_search_docs(saved_search_docs)
+                ),
+            ),
+        )
+
+    packets.append(Packet(placement=Placement(turn_index=turn_index), obj=SectionEnd()))
+
+    return packets

backend/requirements/default.txt

@@ -698,7 +698,7 @@ py-key-value-aio==0.4.4
     # via fastmcp
 pyairtable==3.0.1
     # via onyx
-pyasn1==0.6.3
+pyasn1==0.6.2
     # via
     #   pyasn1-modules
     #   rsa

backend/requirements/dev.txt

@@ -263,7 +263,7 @@ oauthlib==3.2.2
     # via
     #   kubernetes
     #   requests-oauthlib
-onyx-devtools==0.7.1
+onyx-devtools==0.7.0
     # via onyx
 openai==2.14.0
     # via
@@ -326,7 +326,7 @@ pure-eval==0.2.3
     # via stack-data
 py==1.11.0
     # via retry
-pyasn1==0.6.3
+pyasn1==0.6.2
     # via
     #   pyasn1-modules
     #   rsa

backend/requirements/ee.txt

@@ -195,7 +195,7 @@ propcache==0.4.1
     #   yarl
 py==1.11.0
     # via retry
-pyasn1==0.6.3
+pyasn1==0.6.2
     # via
     #   pyasn1-modules
     #   rsa

backend/requirements/model_server.txt

@@ -285,7 +285,7 @@ psutil==7.1.3
     # via accelerate
 py==1.11.0
     # via retry
-pyasn1==0.6.3
+pyasn1==0.6.2
     # via
     #   pyasn1-modules
     #   rsa

backend/scripts/run_industryrag_bench_questions.py

@@ -1,471 +0,0 @@
-from __future__ import annotations
-
-import argparse
-import asyncio
-import json
-import logging
-import sys
-from dataclasses import asdict
-from dataclasses import dataclass
-from pathlib import Path
-from typing import Any
-from typing import TypedDict
-from typing import TypeGuard
-
-import aiohttp
-
-
-logging.basicConfig(
-    level=logging.INFO,
-    format="%(asctime)s %(levelname)s %(message)s",
-)
-logger = logging.getLogger(__name__)
-
-
-DEFAULT_API_BASE = "http://localhost:3000"
-INTERNAL_SEARCH_TOOL_NAME = "internal_search"
-INTERNAL_SEARCH_IN_CODE_TOOL_ID = "SearchTool"
-MAX_REQUEST_ATTEMPTS = 5
-RETRIABLE_STATUS_CODES = {429, 500, 502, 503, 504}
-
-
-@dataclass(frozen=True)
-class QuestionRecord:
-    question_id: str
-    question: str
-
-
-@dataclass(frozen=True)
-class AnswerRecord:
-    question_id: str
-    answer: str
-    document_ids: list[str]
-
-
-@dataclass(frozen=True)
-class FailedQuestionRecord:
-    question_id: str
-    error: str
-
-
-class Citation(TypedDict, total=False):
-    citation_number: int
-    document_id: str
-
-
-def parse_args() -> argparse.Namespace:
-    parser = argparse.ArgumentParser(
-        description=(
-            "Submit questions to Onyx chat with internal search forced and write "
-            "answers to a JSONL file."
-        )
-    )
-    parser.add_argument(
-        "--questions-file",
-        type=Path,
-        required=True,
-        help="Path to the input questions JSONL file.",
-    )
-    parser.add_argument(
-        "--output-file",
-        type=Path,
-        required=True,
-        help="Path to the output answers JSONL file.",
-    )
-    parser.add_argument(
-        "--api-key",
-        type=str,
-        required=True,
-        help="API key used to authenticate against Onyx.",
-    )
-    parser.add_argument(
-        "--api-base",
-        type=str,
-        default=DEFAULT_API_BASE,
-        help=(
-            "Frontend base URL for Onyx. If `/api` is omitted, it will be added "
-            f"automatically. Default: {DEFAULT_API_BASE}"
-        ),
-    )
-    parser.add_argument(
-        "--parallelism",
-        type=int,
-        default=1,
-        help="Number of questions to process in parallel. Default: 1.",
-    )
-    parser.add_argument(
-        "--max-questions",
-        type=int,
-        default=None,
-        help="Optional cap on how many questions to process. Defaults to all.",
-    )
-    return parser.parse_args()
-
-
-def normalize_api_base(api_base: str) -> str:
-    normalized = api_base.rstrip("/")
-    if normalized.endswith("/api"):
-        return normalized
-    return f"{normalized}/api"
-
-
-def load_questions(questions_file: Path) -> list[QuestionRecord]:
-    if not questions_file.exists():
-        raise FileNotFoundError(f"Questions file not found: {questions_file}")
-
-    questions: list[QuestionRecord] = []
-    with questions_file.open("r", encoding="utf-8") as file:
-        for line_number, line in enumerate(file, start=1):
-            stripped_line = line.strip()
-            if not stripped_line:
-                continue
-
-            try:
-                payload = json.loads(stripped_line)
-            except json.JSONDecodeError as exc:
-                raise ValueError(
-                    f"Invalid JSON on line {line_number} of {questions_file}"
-                ) from exc
-
-            question_id = payload.get("question_id")
-            question = payload.get("question")
-
-            if not isinstance(question_id, str) or not question_id:
-                raise ValueError(
-                    f"Line {line_number} is missing a non-empty `question_id`."
-                )
-            if not isinstance(question, str) or not question:
-                raise ValueError(
-                    f"Line {line_number} is missing a non-empty `question`."
-                )
-
-            questions.append(QuestionRecord(question_id=question_id, question=question))
-
-    return questions
-
-
-async def read_json_response(
-    response: aiohttp.ClientResponse,
-) -> dict[str, Any] | list[dict[str, Any]]:
-    response_text = await response.text()
-    if response.status >= 400:
-        raise RuntimeError(
-            f"Request to {response.url} failed with {response.status}: {response_text}"
-        )
-
-    try:
-        payload = json.loads(response_text)
-    except json.JSONDecodeError as exc:
-        raise RuntimeError(
-            f"Request to {response.url} returned non-JSON content: {response_text}"
-        ) from exc
-
-    if not isinstance(payload, (dict, list)):
-        raise RuntimeError(
-            f"Unexpected response payload type from {response.url}: {type(payload)}"
-        )
-
-    return payload
-
-
-async def request_json_with_retries(
-    session: aiohttp.ClientSession,
-    method: str,
-    url: str,
-    headers: dict[str, str],
-    json_payload: dict[str, Any] | None = None,
-) -> dict[str, Any] | list[dict[str, Any]]:
-    backoff_seconds = 1.0
-
-    for attempt in range(1, MAX_REQUEST_ATTEMPTS + 1):
-        try:
-            async with session.request(
-                method=method,
-                url=url,
-                headers=headers,
-                json=json_payload,
-            ) as response:
-                if (
-                    response.status in RETRIABLE_STATUS_CODES
-                    and attempt < MAX_REQUEST_ATTEMPTS
-                ):
-                    response_text = await response.text()
-                    logger.warning(
-                        "Retryable response from %s on attempt %s/%s: %s %s",
-                        url,
-                        attempt,
-                        MAX_REQUEST_ATTEMPTS,
-                        response.status,
-                        response_text,
-                    )
-                    await asyncio.sleep(backoff_seconds)
-                    backoff_seconds *= 2
-                    continue
-
-                return await read_json_response(response)
-        except (aiohttp.ClientError, asyncio.TimeoutError) as exc:
-            if attempt == MAX_REQUEST_ATTEMPTS:
-                raise RuntimeError(
-                    f"Request to {url} failed after {MAX_REQUEST_ATTEMPTS} attempts."
-                ) from exc
-
-            logger.warning(
-                "Request to %s failed on attempt %s/%s: %s",
-                url,
-                attempt,
-                MAX_REQUEST_ATTEMPTS,
-                exc,
-            )
-            await asyncio.sleep(backoff_seconds)
-            backoff_seconds *= 2
-
-    raise RuntimeError(f"Request to {url} failed unexpectedly.")
-
-
-def extract_document_ids(citation_info: object) -> list[str]:
-    if not isinstance(citation_info, list):
-        return []
-
-    sorted_citations = sorted(
-        (citation for citation in citation_info if _is_valid_citation(citation)),
-        key=_citation_sort_key,
-    )
-
-    document_ids: list[str] = []
-    seen_document_ids: set[str] = set()
-    for citation in sorted_citations:
-        document_id = citation["document_id"]
-        if document_id not in seen_document_ids:
-            seen_document_ids.add(document_id)
-            document_ids.append(document_id)
-
-    return document_ids
-
-
-def _is_valid_citation(citation: object) -> TypeGuard[Citation]:
-    return (
-        isinstance(citation, dict)
-        and isinstance(citation.get("document_id"), str)
-        and bool(citation["document_id"])
-    )
-
-
-def _citation_sort_key(citation: Citation) -> int:
-    citation_number = citation.get("citation_number")
-    if isinstance(citation_number, int):
-        return citation_number
-    return sys.maxsize
-
-
-async def fetch_internal_search_tool_id(
-    session: aiohttp.ClientSession,
-    api_base: str,
-    headers: dict[str, str],
-) -> int:
-    payload = await request_json_with_retries(
-        session=session,
-        method="GET",
-        url=f"{api_base}/tool",
-        headers=headers,
-    )
-
-    if not isinstance(payload, list):
-        raise RuntimeError("Expected `/tool` to return a list.")
-
-    for tool in payload:
-        if not isinstance(tool, dict):
-            continue
-
-        if tool.get("in_code_tool_id") == INTERNAL_SEARCH_IN_CODE_TOOL_ID:
-            tool_id = tool.get("id")
-            if isinstance(tool_id, int):
-                return tool_id
-
-    for tool in payload:
-        if not isinstance(tool, dict):
-            continue
-
-        if tool.get("name") == INTERNAL_SEARCH_TOOL_NAME:
-            tool_id = tool.get("id")
-            if isinstance(tool_id, int):
-                return tool_id
-
-    raise RuntimeError(
-        "Could not find the internal search tool in `/tool`. "
-        "Make sure SearchTool is available for this environment."
-    )
-
-
-async def submit_question(
-    session: aiohttp.ClientSession,
-    api_base: str,
-    headers: dict[str, str],
-    internal_search_tool_id: int,
-    question_record: QuestionRecord,
-) -> AnswerRecord:
-    payload = {
-        "message": question_record.question,
-        "chat_session_info": {"persona_id": 0},
-        "parent_message_id": None,
-        "file_descriptors": [],
-        "allowed_tool_ids": [internal_search_tool_id],
-        "forced_tool_id": internal_search_tool_id,
-        "stream": False,
-    }
-
-    response_payload = await request_json_with_retries(
-        session=session,
-        method="POST",
-        url=f"{api_base}/chat/send-chat-message",
-        headers=headers,
-        json_payload=payload,
-    )
-
-    if not isinstance(response_payload, dict):
-        raise RuntimeError(
-            "Expected `/chat/send-chat-message` to return an object when `stream=false`."
-        )
-
-    answer = response_payload.get("answer_citationless")
-    if not isinstance(answer, str):
-        answer = response_payload.get("answer")
-
-    if not isinstance(answer, str):
-        raise RuntimeError(
-            f"Response for question {question_record.question_id} is missing `answer`."
-        )
-
-    return AnswerRecord(
-        question_id=question_record.question_id,
-        answer=answer,
-        document_ids=extract_document_ids(response_payload.get("citation_info")),
-    )
-
-
-async def generate_answers(
-    questions: list[QuestionRecord],
-    output_file: Path,
-    api_base: str,
-    api_key: str,
-    parallelism: int,
-) -> None:
-    if parallelism < 1:
-        raise ValueError("`--parallelism` must be at least 1.")
-
-    headers = {
-        "Authorization": f"Bearer {api_key}",
-        "Content-Type": "application/json",
-    }
-
-    timeout = aiohttp.ClientTimeout(
-        total=None,
-        connect=30,
-        sock_connect=30,
-        sock_read=600,
-    )
-    connector = aiohttp.TCPConnector(limit=parallelism)
-
-    output_file.parent.mkdir(parents=True, exist_ok=True)
-    with output_file.open("a", encoding="utf-8") as file:
-        async with aiohttp.ClientSession(
-            timeout=timeout, connector=connector
-        ) as session:
-            internal_search_tool_id = await fetch_internal_search_tool_id(
-                session=session,
-                api_base=api_base,
-                headers=headers,
-            )
-            logger.info("Using internal search tool id %s", internal_search_tool_id)
-
-            semaphore = asyncio.Semaphore(parallelism)
-            progress_lock = asyncio.Lock()
-            write_lock = asyncio.Lock()
-            completed = 0
-            successful = 0
-            failed_questions: list[FailedQuestionRecord] = []
-            total = len(questions)
-
-            async def process_question(question_record: QuestionRecord) -> None:
-                nonlocal completed
-                nonlocal successful
-
-                try:
-                    async with semaphore:
-                        result = await submit_question(
-                            session=session,
-                            api_base=api_base,
-                            headers=headers,
-                            internal_search_tool_id=internal_search_tool_id,
-                            question_record=question_record,
-                        )
-                except Exception as exc:
-                    async with progress_lock:
-                        completed += 1
-                        failed_questions.append(
-                            FailedQuestionRecord(
-                                question_id=question_record.question_id,
-                                error=str(exc),
-                            )
-                        )
-                        logger.exception(
-                            "Failed question %s (%s/%s)",
-                            question_record.question_id,
-                            completed,
-                            total,
-                        )
-                    return
-
-                async with write_lock:
-                    file.write(json.dumps(asdict(result), ensure_ascii=False))
-                    file.write("\n")
-                    file.flush()
-
-                async with progress_lock:
-                    completed += 1
-                    successful += 1
-                    logger.info("Processed %s/%s questions", completed, total)
-
-            await asyncio.gather(
-                *(process_question(question_record) for question_record in questions)
-            )
-
-            if failed_questions:
-                logger.warning(
-                    "Completed with %s failed questions and %s successful questions.",
-                    len(failed_questions),
-                    successful,
-                )
-                for failed_question in failed_questions:
-                    logger.warning(
-                        "Failed question %s: %s",
-                        failed_question.question_id,
-                        failed_question.error,
-                    )
-
-
-def main() -> None:
-    args = parse_args()
-    questions = load_questions(args.questions_file)
-    api_base = normalize_api_base(args.api_base)
-
-    if args.max_questions is not None:
-        if args.max_questions < 1:
-            raise ValueError("`--max-questions` must be at least 1 when provided.")
-        questions = questions[: args.max_questions]
-
-    logger.info("Loaded %s questions from %s", len(questions), args.questions_file)
-    logger.info("Writing answers to %s", args.output_file)
-
-    asyncio.run(
-        generate_answers(
-            questions=questions,
-            output_file=args.output_file,
-            api_base=api_base,
-            api_key=args.api_key,
-            parallelism=args.parallelism,
-        )
-    )
-
-
-if __name__ == "__main__":
-    main()

backend/scripts/upload_files_as_connectors.py

@@ -1,291 +0,0 @@
-"""
-Script to upload files from a directory as individual file connectors in Onyx.
-Each file gets its own connector named after the file.
-
-Usage:
-    python upload_files_as_connectors.py --data-dir /path/to/files --api-key YOUR_KEY
-    python upload_files_as_connectors.py --data-dir /path/to/files --api-key YOUR_KEY --api-base http://onyxserver:3000
-    python upload_files_as_connectors.py --data-dir /path/to/files --api-key YOUR_KEY --file-glob '*.zip'
-
-Requires:
-    pip install requests
-"""
-
-import argparse
-import fnmatch
-import os
-import sys
-import threading
-import time
-
-import requests
-
-REQUEST_TIMEOUT = 900  # 15 minutes
-
-
-def _elapsed_printer(label: str, stop_event: threading.Event) -> None:
-    """Print a live elapsed-time counter until stop_event is set."""
-    start = time.monotonic()
-    while not stop_event.wait(timeout=1):
-        elapsed = int(time.monotonic() - start)
-        m, s = divmod(elapsed, 60)
-        print(f"\r  {label} ... {m:02d}:{s:02d}", end="", flush=True)
-    elapsed = int(time.monotonic() - start)
-    m, s = divmod(elapsed, 60)
-    print(f"\r  {label} ... {m:02d}:{s:02d} done")
-
-
-def _timed_request(label: str, fn: object) -> requests.Response:
-    """Run a request function while displaying a live elapsed timer."""
-    stop = threading.Event()
-    t = threading.Thread(target=_elapsed_printer, args=(label, stop), daemon=True)
-    t.start()
-    try:
-        resp = fn()  # type: ignore[operator]
-    finally:
-        stop.set()
-        t.join()
-    return resp
-
-
-def upload_file(
-    session: requests.Session, base_url: str, file_path: str
-) -> dict | None:
-    """Upload a single file and return the response with file_paths and file_names."""
-    with open(file_path, "rb") as f:
-        resp = _timed_request(
-            "Uploading",
-            lambda: session.post(
-                f"{base_url}/api/manage/admin/connector/file/upload",
-                files={"files": (os.path.basename(file_path), f)},
-                timeout=REQUEST_TIMEOUT,
-            ),
-        )
-    if not resp.ok:
-        print(f"  ERROR uploading: {resp.text}")
-        return None
-    return resp.json()
-
-
-def create_connector(
-    session: requests.Session,
-    base_url: str,
-    name: str,
-    file_paths: list[str],
-    file_names: list[str],
-    zip_metadata_file_id: str | None,
-) -> int | None:
-    """Create a file connector and return its ID."""
-    resp = _timed_request(
-        "Creating connector",
-        lambda: session.post(
-            f"{base_url}/api/manage/admin/connector",
-            json={
-                "name": name,
-                "source": "file",
-                "input_type": "load_state",
-                "connector_specific_config": {
-                    "file_locations": file_paths,
-                    "file_names": file_names,
-                    "zip_metadata_file_id": zip_metadata_file_id,
-                },
-                "refresh_freq": None,
-                "prune_freq": None,
-                "indexing_start": None,
-                "access_type": "public",
-                "groups": [],
-            },
-            timeout=REQUEST_TIMEOUT,
-        ),
-    )
-    if not resp.ok:
-        print(f"  ERROR creating connector: {resp.text}")
-        return None
-    return resp.json()["id"]
-
-
-def create_credential(
-    session: requests.Session, base_url: str, name: str
-) -> int | None:
-    """Create a dummy credential for the file connector."""
-    resp = session.post(
-        f"{base_url}/api/manage/credential",
-        json={
-            "credential_json": {},
-            "admin_public": True,
-            "source": "file",
-            "curator_public": True,
-            "groups": [],
-            "name": name,
-        },
-        timeout=REQUEST_TIMEOUT,
-    )
-    if not resp.ok:
-        print(f"  ERROR creating credential: {resp.text}")
-        return None
-    return resp.json()["id"]
-
-
-def link_credential(
-    session: requests.Session,
-    base_url: str,
-    connector_id: int,
-    credential_id: int,
-    name: str,
-) -> bool:
-    """Link the connector to the credential (create CC pair)."""
-    resp = session.put(
-        f"{base_url}/api/manage/connector/{connector_id}/credential/{credential_id}",
-        json={
-            "name": name,
-            "access_type": "public",
-            "groups": [],
-            "auto_sync_options": None,
-            "processing_mode": "REGULAR",
-        },
-        timeout=REQUEST_TIMEOUT,
-    )
-    if not resp.ok:
-        print(f"  ERROR linking credential: {resp.text}")
-        return False
-    return True
-
-
-def run_connector(
-    session: requests.Session,
-    base_url: str,
-    connector_id: int,
-    credential_id: int,
-) -> bool:
-    """Trigger the connector to start indexing."""
-    resp = session.post(
-        f"{base_url}/api/manage/admin/connector/run-once",
-        json={
-            "connector_id": connector_id,
-            "credentialIds": [credential_id],
-            "from_beginning": False,
-        },
-        timeout=REQUEST_TIMEOUT,
-    )
-    if not resp.ok:
-        print(f"  ERROR running connector: {resp.text}")
-        return False
-    return True
-
-
-def process_file(session: requests.Session, base_url: str, file_path: str) -> bool:
-    """Process a single file through the full connector creation flow."""
-    file_name = os.path.basename(file_path)
-    connector_name = file_name
-    print(f"Processing: {file_name}")
-
-    # Step 1: Upload
-    upload_resp = upload_file(session, base_url, file_path)
-    if not upload_resp:
-        return False
-
-    # Step 2: Create connector
-    connector_id = create_connector(
-        session,
-        base_url,
-        name=f"FileConnector-{connector_name}",
-        file_paths=upload_resp["file_paths"],
-        file_names=upload_resp["file_names"],
-        zip_metadata_file_id=upload_resp.get("zip_metadata_file_id"),
-    )
-    if connector_id is None:
-        return False
-
-    # Step 3: Create credential
-    credential_id = create_credential(session, base_url, name=connector_name)
-    if credential_id is None:
-        return False
-
-    # Step 4: Link connector to credential
-    if not link_credential(
-        session, base_url, connector_id, credential_id, connector_name
-    ):
-        return False
-
-    # Step 5: Trigger indexing
-    if not run_connector(session, base_url, connector_id, credential_id):
-        return False
-
-    print(f"  OK (connector_id={connector_id})")
-    return True
-
-
-def get_authenticated_session(api_key: str) -> requests.Session:
-    """Create a session authenticated with an API key."""
-    session = requests.Session()
-    session.headers.update({"Authorization": f"Bearer {api_key}"})
-    return session
-
-
-def main() -> None:
-    parser = argparse.ArgumentParser(
-        description="Upload files as individual Onyx file connectors."
-    )
-    parser.add_argument(
-        "--data-dir",
-        required=True,
-        help="Directory containing files to upload.",
-    )
-    parser.add_argument(
-        "--api-base",
-        default="http://localhost:3000",
-        help="Base URL for the Onyx API (default: http://localhost:3000).",
-    )
-    parser.add_argument(
-        "--api-key",
-        required=True,
-        help="API key for authentication.",
-    )
-    parser.add_argument(
-        "--file-glob",
-        default=None,
-        help="Glob pattern to filter files (e.g. '*.json', '*.zip').",
-    )
-    args = parser.parse_args()
-
-    data_dir = args.data_dir
-    base_url = args.api_base.rstrip("/")
-    api_key = args.api_key
-    file_glob = args.file_glob
-
-    if not os.path.isdir(data_dir):
-        print(f"Error: {data_dir} is not a directory")
-        sys.exit(1)
-
-    script_path = os.path.realpath(__file__)
-    files = sorted(
-        os.path.join(data_dir, f)
-        for f in os.listdir(data_dir)
-        if os.path.isfile(os.path.join(data_dir, f))
-        and os.path.realpath(os.path.join(data_dir, f)) != script_path
-        and (file_glob is None or fnmatch.fnmatch(f, file_glob))
-    )
-
-    if not files:
-        print(f"No files found in {data_dir}")
-        sys.exit(1)
-
-    print(f"Found {len(files)} file(s) in {data_dir}\n")
-
-    session = get_authenticated_session(api_key)
-
-    success = 0
-    failed = 0
-    for file_path in files:
-        if process_file(session, base_url, file_path):
-            success += 1
-        else:
-            failed += 1
-        # Small delay to avoid overwhelming the server
-        time.sleep(0.5)
-
-    print(f"\nDone: {success} succeeded, {failed} failed out of {len(files)} files.")
-
-
-if __name__ == "__main__":
-    main()

backend/tests/external_dependency_unit/document_index/test_document_index_old.py

@@ -297,10 +297,6 @@ def index_batch_params(
 class TestDocumentIndexOld:
     """Tests the old DocumentIndex interface."""
 
-    # TODO(ENG-3864)(andrei): Re-enable this test.
-    @pytest.mark.xfail(
-        reason="Flaky test: Retrieved chunks vary non-deterministically before and after changing user projects and personas. Likely a timing issue with the index being updated."
-    )
     def test_update_single_can_clear_user_projects_and_personas(
         self,
         document_indices: list[DocumentIndex],

backend/tests/external_dependency_unit/opensearch/test_opensearch_client.py

@@ -22,8 +22,6 @@ from onyx.document_index.interfaces_new import TenantState
 from onyx.document_index.opensearch.client import OpenSearchIndexClient
 from onyx.document_index.opensearch.client import wait_for_opensearch_with_timeout
 from onyx.document_index.opensearch.constants import DEFAULT_MAX_CHUNK_SIZE
-from onyx.document_index.opensearch.constants import HybridSearchNormalizationPipeline
-from onyx.document_index.opensearch.constants import HybridSearchSubqueryConfiguration
 from onyx.document_index.opensearch.opensearch_document_index import (
     generate_opensearch_filtered_access_control_list,
 )
@@ -33,14 +31,9 @@ from onyx.document_index.opensearch.schema import DocumentSchema
 from onyx.document_index.opensearch.schema import get_opensearch_doc_chunk_id
 from onyx.document_index.opensearch.search import DocumentQuery
 from onyx.document_index.opensearch.search import (
-    get_min_max_normalization_pipeline_name_and_config,
-)
-from onyx.document_index.opensearch.search import (
-    get_normalization_pipeline_name_and_config,
-)
-from onyx.document_index.opensearch.search import (
-    get_zscore_normalization_pipeline_name_and_config,
+    MIN_MAX_NORMALIZATION_PIPELINE_CONFIG,
 )
+from onyx.document_index.opensearch.search import MIN_MAX_NORMALIZATION_PIPELINE_NAME
 from shared_configs.configs import POSTGRES_DEFAULT_SCHEMA
 
 
@@ -56,46 +49,6 @@ def _patch_global_tenant_state(monkeypatch: pytest.MonkeyPatch, state: bool) ->
     monkeypatch.setattr("onyx.document_index.opensearch.schema.MULTI_TENANT", state)
 
 
-def _patch_hybrid_search_subquery_configuration(
-    monkeypatch: pytest.MonkeyPatch, configuration: HybridSearchSubqueryConfiguration
-) -> None:
-    """
-    Patches HYBRID_SEARCH_SUBQUERY_CONFIGURATION wherever necessary for this
-    test file.
-
-    Args:
-        monkeypatch: The test instance's monkeypatch instance, used for
-            patching.
-        configuration: The intended state of
-            HYBRID_SEARCH_SUBQUERY_CONFIGURATION.
-    """
-    monkeypatch.setattr(
-        "onyx.document_index.opensearch.constants.HYBRID_SEARCH_SUBQUERY_CONFIGURATION",
-        configuration,
-    )
-    monkeypatch.setattr(
-        "onyx.document_index.opensearch.search.HYBRID_SEARCH_SUBQUERY_CONFIGURATION",
-        configuration,
-    )
-
-
-def _patch_hybrid_search_normalization_pipeline(
-    monkeypatch: pytest.MonkeyPatch, pipeline: HybridSearchNormalizationPipeline
-) -> None:
-    """
-    Patches HYBRID_SEARCH_NORMALIZATION_PIPELINE wherever necessary for this
-    test file.
-    """
-    monkeypatch.setattr(
-        "onyx.document_index.opensearch.constants.HYBRID_SEARCH_NORMALIZATION_PIPELINE",
-        pipeline,
-    )
-    monkeypatch.setattr(
-        "onyx.document_index.opensearch.search.HYBRID_SEARCH_NORMALIZATION_PIPELINE",
-        pipeline,
-    )
-
-
 def _create_test_document_chunk(
     document_id: str,
     content: str,
@@ -191,27 +144,14 @@ def test_client(
 @pytest.fixture(scope="function")
 def search_pipeline(test_client: OpenSearchIndexClient) -> Generator[None, None, None]:
     """Creates a search pipeline for testing with automatic cleanup."""
-    min_max_normalization_pipeline_name, min_max_normalization_pipeline_config = (
-        get_min_max_normalization_pipeline_name_and_config()
-    )
-    zscore_normalization_pipeline_name, zscore_normalization_pipeline_config = (
-        get_zscore_normalization_pipeline_name_and_config()
-    )
     test_client.create_search_pipeline(
-        pipeline_id=min_max_normalization_pipeline_name,
-        pipeline_body=min_max_normalization_pipeline_config,
-    )
-    test_client.create_search_pipeline(
-        pipeline_id=zscore_normalization_pipeline_name,
-        pipeline_body=zscore_normalization_pipeline_config,
+        pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
+        pipeline_body=MIN_MAX_NORMALIZATION_PIPELINE_CONFIG,
     )
     yield  # Test runs here.
     try:
         test_client.delete_search_pipeline(
-            pipeline_id=min_max_normalization_pipeline_name,
-        )
-        test_client.delete_search_pipeline(
-            pipeline_id=zscore_normalization_pipeline_name,
+            pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
         )
     except Exception:
         pass
@@ -437,19 +377,18 @@ class TestOpenSearchClient:
         self, test_client: OpenSearchIndexClient
     ) -> None:
         """Tests creating and deleting a search pipeline."""
-        # Precondition.
-        pipeline_name, pipeline_config = get_normalization_pipeline_name_and_config()
-
         # Under test and postcondition.
         # Should not raise.
         test_client.create_search_pipeline(
-            pipeline_id=pipeline_name,
-            pipeline_body=pipeline_config,
+            pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
+            pipeline_body=MIN_MAX_NORMALIZATION_PIPELINE_CONFIG,
         )
 
         # Under test and postcondition.
         # Should not raise.
-        test_client.delete_search_pipeline(pipeline_id=pipeline_name)
+        test_client.delete_search_pipeline(
+            pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME
+        )
 
     def test_index_document(
         self, test_client: OpenSearchIndexClient, monkeypatch: pytest.MonkeyPatch
@@ -795,13 +734,13 @@ class TestOpenSearchClient:
                 properties_to_update={"hidden": True},
             )
 
-    def test_hybrid_search_configurations_and_pipelines(
+    def test_hybrid_search_with_pipeline(
         self,
         test_client: OpenSearchIndexClient,
         search_pipeline: None,  # noqa: ARG002
         monkeypatch: pytest.MonkeyPatch,
     ) -> None:
-        """Tests all hybrid search configurations and pipelines."""
+        """Tests hybrid search with a normalization pipeline."""
         # Precondition.
         _patch_global_tenant_state(monkeypatch, False)
         tenant_state = TenantState(tenant_id=POSTGRES_DEFAULT_SCHEMA, multitenant=False)
@@ -810,6 +749,7 @@ class TestOpenSearchClient:
         )
         settings = DocumentSchema.get_index_settings()
         test_client.create_index(mappings=mappings, settings=settings)
+
         # Index documents.
         docs = {
             "doc-1": _create_test_document_chunk(
@@ -840,58 +780,40 @@ class TestOpenSearchClient:
         # Refresh index to make documents searchable.
         test_client.refresh_index()
 
-        for configuration in HybridSearchSubqueryConfiguration:
-            _patch_hybrid_search_subquery_configuration(monkeypatch, configuration)
-            for pipeline in HybridSearchNormalizationPipeline:
-                _patch_hybrid_search_normalization_pipeline(monkeypatch, pipeline)
-                pipeline_name, pipeline_config = (
-                    get_normalization_pipeline_name_and_config()
-                )
-                test_client.create_search_pipeline(
-                    pipeline_id=pipeline_name,
-                    pipeline_body=pipeline_config,
-                )
+        # Search query.
+        query_text = "Python programming"
+        query_vector = _generate_test_vector(0.12)
+        search_body = DocumentQuery.get_hybrid_search_query(
+            query_text=query_text,
+            query_vector=query_vector,
+            num_hits=5,
+            tenant_state=tenant_state,
+            # We're not worried about filtering here. tenant_id in this object
+            # is not relevant.
+            index_filters=IndexFilters(access_control_list=None, tenant_id=None),
+            include_hidden=False,
+        )
 
-                # Search query.
-                query_text = "Python programming"
-                query_vector = _generate_test_vector(0.12)
-                search_body = DocumentQuery.get_hybrid_search_query(
-                    query_text=query_text,
-                    query_vector=query_vector,
-                    num_hits=5,
-                    tenant_state=tenant_state,
-                    # We're not worried about filtering here. tenant_id in this object
-                    # is not relevant.
-                    index_filters=IndexFilters(
-                        access_control_list=None, tenant_id=None
-                    ),
-                    include_hidden=False,
-                )
+        # Under test.
+        results = test_client.search(
+            body=search_body, search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME
+        )
 
-                # Under test.
-                results = test_client.search(
-                    body=search_body, search_pipeline_id=pipeline_name
-                )
-
-                # Postcondition.
-                assert len(results) == len(docs)
-                # Assert that all the chunks above are present.
-                assert all(
-                    chunk.document_chunk.document_id in docs.keys() for chunk in results
-                )
-                # Make sure the chunk contents are preserved.
-                for i, chunk in enumerate(results):
-                    assert (
-                        chunk.document_chunk == docs[chunk.document_chunk.document_id]
-                    )
-                    # Make sure score reporting seems reasonable (it should not be None
-                    # or 0).
-                    assert chunk.score
-                    # Make sure there is some kind of match highlight only for the first
-                    # result. The other results are so bad they're not expected to have
-                    # match highlights.
-                    if i == 0:
-                        assert chunk.match_highlights.get(CONTENT_FIELD_NAME, [])
+        # Postcondition.
+        assert len(results) == len(docs)
+        # Assert that all the chunks above are present.
+        assert all(chunk.document_chunk.document_id in docs.keys() for chunk in results)
+        # Make sure the chunk contents are preserved.
+        for i, chunk in enumerate(results):
+            assert chunk.document_chunk == docs[chunk.document_chunk.document_id]
+            # Make sure score reporting seems reasonable (it should not be None
+            # or 0).
+            assert chunk.score
+            # Make sure there is some kind of match highlight only for the first
+            # result. The other results are so bad they're not expected to have
+            # match highlights.
+            if i == 0:
+                assert chunk.match_highlights.get(CONTENT_FIELD_NAME, [])
 
     def test_search_empty_index(
         self,
@@ -923,10 +845,11 @@ class TestOpenSearchClient:
             index_filters=IndexFilters(access_control_list=None, tenant_id=None),
             include_hidden=False,
         )
-        pipeline_name, _ = get_normalization_pipeline_name_and_config()
 
         # Under test.
-        results = test_client.search(body=search_body, search_pipeline_id=pipeline_name)
+        results = test_client.search(
+            body=search_body, search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME
+        )
 
         # Postcondition.
         assert len(results) == 0
@@ -1025,10 +948,11 @@ class TestOpenSearchClient:
             ),
             include_hidden=False,
         )
-        pipeline_name, _ = get_normalization_pipeline_name_and_config()
 
         # Under test.
-        results = test_client.search(body=search_body, search_pipeline_id=pipeline_name)
+        results = test_client.search(
+            body=search_body, search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME
+        )
 
         # Postcondition.
         # Should only get the public, non-hidden document, and the private
@@ -1143,10 +1067,11 @@ class TestOpenSearchClient:
             index_filters=IndexFilters(access_control_list=[], tenant_id=None),
             include_hidden=False,
         )
-        pipeline_name, _ = get_normalization_pipeline_name_and_config()
 
         # Under test.
-        results = test_client.search(body=search_body, search_pipeline_id=pipeline_name)
+        results = test_client.search(
+            body=search_body, search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME
+        )
 
         # Postcondition.
         # Should only get public, non-hidden documents (3 out of 5).
@@ -1516,16 +1441,15 @@ class TestOpenSearchClient:
             ),
             include_hidden=False,
         )
-        pipeline_name, _ = get_normalization_pipeline_name_and_config()
 
         # Under test.
         last_week_results = test_client.search(
             body=last_week_search_body,
-            search_pipeline_id=pipeline_name,
+            search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
         )
         last_six_months_results = test_client.search(
             body=last_six_months_search_body,
-            search_pipeline_id=pipeline_name,
+            search_pipeline_id=MIN_MAX_NORMALIZATION_PIPELINE_NAME,
         )
 
         # Postcondition.

backend/tests/unit/onyx/chat/test_llm_loop.py

@@ -1,13 +1,9 @@
-"""Tests for llm_loop.py, including history construction and empty-response paths."""
-
-from unittest.mock import Mock
+"""Tests for llm_loop.py, specifically the construct_message_history function."""
 
 import pytest
 
-from onyx.chat.llm_loop import _build_empty_llm_response_error
 from onyx.chat.llm_loop import _try_fallback_tool_extraction
 from onyx.chat.llm_loop import construct_message_history
-from onyx.chat.llm_loop import EmptyLLMResponseError
 from onyx.chat.models import ChatLoadedFile
 from onyx.chat.models import ChatMessageSimple
 from onyx.chat.models import ContextFileMetadata
@@ -17,7 +13,6 @@ from onyx.chat.models import LlmStepResult
 from onyx.chat.models import ToolCallSimple
 from onyx.configs.constants import MessageType
 from onyx.file_store.models import ChatFileType
-from onyx.llm.interfaces import LLMConfig
 from onyx.llm.interfaces import ToolChoiceOptions
 from onyx.server.query_and_chat.placement import Placement
 from onyx.tools.models import ToolCallKickoff
@@ -1172,57 +1167,3 @@ class TestFallbackToolExtraction:
 
         assert result is llm_step_result
         assert attempted is False
-
-
-class TestEmptyLlmResponseClassification:
-    def _make_llm(self, provider: str = "openai", model: str = "gpt-5.2") -> Mock:
-        llm = Mock()
-        llm.config = LLMConfig(
-            model_provider=provider,
-            model_name=model,
-            temperature=0.0,
-            max_input_tokens=4096,
-        )
-        return llm
-
-    def test_openai_empty_stream_is_classified_as_budget_exceeded(
-        self, monkeypatch: pytest.MonkeyPatch
-    ) -> None:
-        monkeypatch.setattr("onyx.chat.llm_loop.is_true_openai_model", lambda *_: True)
-
-        err = _build_empty_llm_response_error(
-            llm=self._make_llm(),
-            llm_step_result=LlmStepResult(
-                reasoning=None,
-                answer=None,
-                tool_calls=None,
-                raw_answer=None,
-            ),
-            tool_choice=ToolChoiceOptions.AUTO,
-        )
-
-        assert isinstance(err, EmptyLLMResponseError)
-        assert err.error_code == "BUDGET_EXCEEDED"
-        assert err.is_retryable is False
-        assert "quota" in err.client_error_msg.lower()
-
-    def test_reasoning_only_response_uses_generic_empty_response_error(
-        self, monkeypatch: pytest.MonkeyPatch
-    ) -> None:
-        monkeypatch.setattr("onyx.chat.llm_loop.is_true_openai_model", lambda *_: True)
-
-        err = _build_empty_llm_response_error(
-            llm=self._make_llm(),
-            llm_step_result=LlmStepResult(
-                reasoning="scratchpad only",
-                answer=None,
-                tool_calls=None,
-                raw_answer=None,
-            ),
-            tool_choice=ToolChoiceOptions.AUTO,
-        )
-
-        assert isinstance(err, EmptyLLMResponseError)
-        assert err.error_code == "EMPTY_LLM_RESPONSE"
-        assert err.is_retryable is True
-        assert "quota" not in err.client_error_msg.lower()

backend/tests/unit/onyx/chat/test_process_message.py

@@ -1,34 +0,0 @@
-from onyx.chat.process_message import remove_answer_citations
-
-
-def test_remove_answer_citations_strips_http_markdown_citation() -> None:
-    answer = "The answer is Paris [[1]](https://example.com/doc)."
-
-    assert remove_answer_citations(answer) == "The answer is Paris."
-
-
-def test_remove_answer_citations_strips_empty_markdown_citation() -> None:
-    answer = "The answer is Paris [[1]]()."
-
-    assert remove_answer_citations(answer) == "The answer is Paris."
-
-
-def test_remove_answer_citations_strips_citation_with_parentheses_in_url() -> None:
-    answer = (
-        "The answer is Paris "
-        "[[1]](https://en.wikipedia.org/wiki/Function_(mathematics))."
-    )
-
-    assert remove_answer_citations(answer) == "The answer is Paris."
-
-
-def test_remove_answer_citations_preserves_non_citation_markdown_links() -> None:
-    answer = (
-        "See [reference](https://example.com/Function_(mathematics)) "
-        "for context [[1]](https://en.wikipedia.org/wiki/Function_(mathematics))."
-    )
-
-    assert (
-        remove_answer_citations(answer)
-        == "See [reference](https://example.com/Function_(mathematics)) for context."
-    )

backend/tests/unit/onyx/chat/test_process_message_mock_llm.py

@@ -3,10 +3,7 @@ from unittest.mock import Mock
 import pytest
 
 from onyx.chat import process_message
-from onyx.chat.models import AnswerStream
-from onyx.chat.models import StreamingError
 from onyx.configs import app_configs
-from onyx.server.query_and_chat.models import MessageResponseIDInfo
 from onyx.server.query_and_chat.models import SendMessageRequest
 
 
@@ -38,26 +35,3 @@ def test_mock_llm_response_requires_integration_mode() -> None:
                 db_session=Mock(),
             )
         )
-
-
-def test_gather_stream_returns_empty_answer_when_streaming_error_only() -> None:
-    packets: AnswerStream = iter(
-        [
-            MessageResponseIDInfo(
-                user_message_id=None,
-                reserved_assistant_message_id=42,
-            ),
-            StreamingError(
-                error="OpenAI quota exceeded",
-                error_code="BUDGET_EXCEEDED",
-                is_retryable=False,
-            ),
-        ]
-    )
-
-    result = process_message.gather_stream(packets)
-
-    assert result.answer == ""
-    assert result.answer_citationless == ""
-    assert result.error_msg == "OpenAI quota exceeded"
-    assert result.message_id == 42

backend/tests/unit/onyx/utils/test_vespa_query.py

@@ -44,13 +44,13 @@ class TestBuildVespaFilters:
         assert result == f'({SOURCE_TYPE} contains "web") and '
 
     def test_acl(self) -> None:
-        """Test with acls — uses weightedSet operator for efficient matching."""
+        """Test with acls."""
         # Single ACL
         filters = IndexFilters(access_control_list=["user1"])
         result = build_vespa_filters(filters)
         assert (
             result
-            == f'!({HIDDEN}=true) and weightedSet(access_control_list, {{"user1":1}}) and '
+            == f'!({HIDDEN}=true) and (access_control_list contains "user1") and '
         )
 
         # Multiple ACL's
@@ -58,7 +58,7 @@ class TestBuildVespaFilters:
         result = build_vespa_filters(filters)
         assert (
             result
-            == f'!({HIDDEN}=true) and weightedSet(access_control_list, {{"user2":1, "group2":1}}) and '
+            == f'!({HIDDEN}=true) and (access_control_list contains "user2" or access_control_list contains "group2") and '
         )
 
     def test_tenant_filter(self) -> None:
@@ -250,7 +250,7 @@ class TestBuildVespaFilters:
         result = build_vespa_filters(filters)
 
         expected = f"!({HIDDEN}=true) and "
-        expected += 'weightedSet(access_control_list, {"user1":1, "group1":1}) and '
+        expected += '(access_control_list contains "user1" or access_control_list contains "group1") and '
         expected += f'({SOURCE_TYPE} contains "web") and '
         expected += f'({METADATA_LIST} contains "color{INDEX_SEPARATOR}red") and '
         # Knowledge scope filters are OR'd together
@@ -290,38 +290,6 @@ class TestBuildVespaFilters:
         expected = f'!({HIDDEN}=true) and (({DOCUMENT_SETS} contains "engineering") or ({DOCUMENT_ID} contains "{str(id1)}")) and '
         assert expected == result
 
-    def test_acl_large_list_uses_weighted_set(self) -> None:
-        """Verify that large ACL lists produce a weightedSet clause
-        instead of OR-chained contains — this is what prevents Vespa
-        HTTP 400 errors for users with thousands of permission groups."""
-        acl = [f"external_group:google_drive_{i}" for i in range(10_000)]
-        acl += ["user_email:user@example.com", "__PUBLIC__"]
-        filters = IndexFilters(access_control_list=acl)
-        result = build_vespa_filters(filters)
-
-        assert "weightedSet(access_control_list, {" in result
-        # Must NOT contain OR-chained contains clauses
-        assert "access_control_list contains" not in result
-        # All entries should be present
-        assert '"external_group:google_drive_0":1' in result
-        assert '"external_group:google_drive_9999":1' in result
-        assert '"user_email:user@example.com":1' in result
-        assert '"__PUBLIC__":1' in result
-
-    def test_acl_empty_strings_filtered(self) -> None:
-        """Empty strings in the ACL list should be filtered out."""
-        filters = IndexFilters(access_control_list=["user1", "", "group1"])
-        result = build_vespa_filters(filters)
-        assert (
-            result
-            == f'!({HIDDEN}=true) and weightedSet(access_control_list, {{"user1":1, "group1":1}}) and '
-        )
-
-        # All empty
-        filters = IndexFilters(access_control_list=["", ""])
-        result = build_vespa_filters(filters)
-        assert result == f"!({HIDDEN}=true) and "
-
     def test_empty_or_none_values(self) -> None:
         """Test with empty or None values in filter lists."""
         # Empty strings in document set

deployment/docker_compose/docker-compose.dev.yml

@@ -12,11 +12,6 @@ services:
   api_server:
     ports:
       - "8080:8080"
-    deploy:
-      resources:
-        limits:
-          cpus: "${API_SERVER_CPU_LIMIT:-0}"
-          memory: "${API_SERVER_MEM_LIMIT:-0}"
 
   # Uncomment the block below to enable the MCP server for Onyx.
   # mcp_server:

deployment/helm/README.md

@@ -19,13 +19,12 @@
 
 ## Output template to file and inspect
 * cd charts/onyx
-* helm template test-output . --set auth.opensearch.values.opensearch_admin_password='StrongPassword123!' > test-output.yaml
+* helm template test-output . > test-output.yaml
 
 ## Test the entire cluster manually
 * cd charts/onyx
-* helm install onyx . -n onyx --set postgresql.primary.persistence.enabled=false --set auth.opensearch.values.opensearch_admin_password='StrongPassword123!'
+* helm install onyx . -n onyx --set postgresql.primary.persistence.enabled=false
   * the postgres flag is to keep the storage ephemeral for testing. You probably don't want to set that in prod.
-  * the OpenSearch admin password must be set on first install unless you are supplying `auth.opensearch.existingSecret`.
   * no flag for ephemeral vespa storage yet, might be good for testing
 * kubectl -n onyx port-forward service/onyx-nginx 8080:80
   * this will forward the local port 8080 to the installed chart for you to run tests, etc.

deployment/helm/charts/onyx/Chart.yaml

@@ -5,7 +5,7 @@ home: https://www.onyx.app/
 sources:
   - "https://github.com/onyx-dot-app/onyx"
 type: application
-version: 0.4.36
+version: 0.4.35
 appVersion: latest
 annotations:
   category: Productivity

deployment/helm/charts/onyx/ci/ct-values.yaml

@@ -1,9 +1,6 @@
 # Values for chart-testing (ct lint/install)
 # This file is automatically used by ct when running lint and install commands
 auth:
-  opensearch:
-    values:
-      opensearch_admin_password: "placeholder-OpenSearch1!"
   userauth:
     values:
       user_auth_secret: "placeholder-for-ci-testing"

deployment/helm/charts/onyx/values.yaml

@@ -1177,17 +1177,12 @@ auth:
     # Secrets values IF existingSecret is empty. Key here must match the value
     # in secretKeys to be used. Values will be base64 encoded in the k8s
     # cluster.
-    # For the bundled OpenSearch chart, the admin password is consumed during
-    # initial cluster setup. Changing this value later will update Onyx's
-    # client credentials, but will not rotate the OpenSearch admin password.
-    # Set this before first install or use existingSecret to preserve the
-    # current secret on upgrade.
     # Password must meet OpenSearch complexity requirements:
     # min 8 chars, uppercase, lowercase, digit, and special character.
-    # Required when auth.opensearch.enabled=true and no existing secret exists.
+    # CHANGE THIS FOR PRODUCTION.
     values:
       opensearch_admin_username: "admin"
-      opensearch_admin_password: ""
+      opensearch_admin_password: "OnyxDev1!"
   userauth:
     # -- Used for password reset / verification tokens and OAuth/OIDC state signing.
     # Disabled by default to preserve upgrade compatibility for existing Helm customers.

pyproject.toml

@@ -144,7 +144,7 @@ dev = [
     "matplotlib==3.10.8",
     "mypy-extensions==1.0.0",
     "mypy==1.13.0",
-    "onyx-devtools==0.7.1",
+    "onyx-devtools==0.7.0",
     "openapi-generator-cli==7.17.0",
     "pandas-stubs~=2.3.3",
     "pre-commit==3.2.2",

tools/ods/cmd/backend.go

@@ -4,11 +4,9 @@ import (
 	"bufio"
 	"errors"
 	"fmt"
-	"net"
 	"os"
 	"os/exec"
 	"path/filepath"
-	"strconv"
 	"strings"
 
 	log "github.com/sirupsen/logrus"
@@ -94,57 +92,12 @@ Examples:
 	return cmd
 }
 
-func isPortAvailable(port int) bool {
-	ln, err := net.Listen("tcp", fmt.Sprintf(":%d", port))
-	if err != nil {
-		return false
-	}
-	_ = ln.Close()
-	return true
-}
-
-func getProcessOnPort(port int) string {
-	out, err := exec.Command("lsof", "-i", fmt.Sprintf(":%d", port), "-t").Output()
-	if err != nil || len(strings.TrimSpace(string(out))) == 0 {
-		return "an unknown process"
-	}
-	pid := strings.Split(strings.TrimSpace(string(out)), "\n")[0]
-	nameOut, err := exec.Command("ps", "-p", pid, "-o", "comm=").Output()
-	if err != nil || len(strings.TrimSpace(string(nameOut))) == 0 {
-		return fmt.Sprintf("process (PID %s)", pid)
-	}
-	return fmt.Sprintf("%s (PID %s)", strings.TrimSpace(string(nameOut)), pid)
-}
-
-func resolvePort(port string) string {
-	portNum, err := strconv.Atoi(port)
-	if err != nil {
-		log.Fatalf("Invalid port %q: %v", port, err)
-	}
-	if isPortAvailable(portNum) {
-		return port
-	}
-	proc := getProcessOnPort(portNum)
-	candidate := portNum + 1
-	for candidate <= 65535 {
-		if isPortAvailable(candidate) {
-			log.Warnf("⚠ Port %d is in use by %s, using available port %d instead.", portNum, proc, candidate)
-			return strconv.Itoa(candidate)
-		}
-		candidate++
-	}
-	log.Fatalf("No available ports found starting from %d", portNum)
-	return port
-}
-
 func runBackendService(name, module, port string, opts *BackendOptions) {
 	root, err := paths.GitRoot()
 	if err != nil {
 		log.Fatalf("Failed to find git root: %v", err)
 	}
 
-	port = resolvePort(port)
-
 	envFile := ensureBackendEnvFile(root)
 	fileVars := loadBackendEnvFile(envFile)
 

uv.lock

@@ -4458,7 +4458,7 @@ requires-dist = [
     { name = "numpy", marker = "extra == 'model-server'", specifier = "==2.4.1" },
     { name = "oauthlib", marker = "extra == 'backend'", specifier = "==3.2.2" },
     { name = "office365-rest-python-client", marker = "extra == 'backend'", specifier = "==2.6.2" },
-    { name = "onyx-devtools", marker = "extra == 'dev'", specifier = "==0.7.1" },
+    { name = "onyx-devtools", marker = "extra == 'dev'", specifier = "==0.7.0" },
     { name = "openai", specifier = "==2.14.0" },
     { name = "openapi-generator-cli", marker = "extra == 'dev'", specifier = "==7.17.0" },
     { name = "openinference-instrumentation", marker = "extra == 'backend'", specifier = "==0.1.42" },
@@ -4563,19 +4563,19 @@ requires-dist = [{ name = "onyx", extras = ["backend", "dev", "ee"], editable =
 
 [[package]]
 name = "onyx-devtools"
-version = "0.7.1"
+version = "0.7.0"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "fastapi" },
     { name = "openapi-generator-cli" },
 ]
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/65/9d/74bcd02583706bdf90c8ac9084eb60bd71d0671392152410ab21b7b68ea1/onyx_devtools-0.7.1-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:178385dce0b413fd2a1f761055a99f556ec536ef5c32963fc273e751813621eb", size = 4007974, upload-time = "2026-03-17T21:10:39.267Z" },
-    { url = "https://files.pythonhosted.org/packages/f0/f8/d8ddb32120428c083c60eb07244479da6e07eaebd31847658a049ab33815/onyx_devtools-0.7.1-py3-none-macosx_11_0_arm64.whl", hash = "sha256:7960ae6e440ebf1584e02d9e1d0c9ef543b1d54c2584cdcace15695aec3121b2", size = 3696924, upload-time = "2026-03-17T21:10:50.716Z" },
-    { url = "https://files.pythonhosted.org/packages/87/21/1e427280066db42ff9dd5f34c70b9dca5d9781f96d0d9a88aaa454fdb432/onyx_devtools-0.7.1-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:6785dda88ca0a3d8464a9bfab76a253ed90da89d53a9c4a67227980f37df1ccf", size = 3568300, upload-time = "2026-03-17T21:10:41.997Z" },
-    { url = "https://files.pythonhosted.org/packages/0e/0e/afbbe1164b3d016ddb5352353cb2541eef5a8b2c04e8f02d5d1319cb8b8c/onyx_devtools-0.7.1-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:9e77f2b725c0c00061a3dda5eba199404b51638cec0bf54fc7611fee1f26db34", size = 3974668, upload-time = "2026-03-17T21:10:43.879Z" },
-    { url = "https://files.pythonhosted.org/packages/8a/a5/22840643289ef4ca83931b7a79fba8f1db7e626b4b870d4b4f8206c4ff5f/onyx_devtools-0.7.1-py3-none-win_amd64.whl", hash = "sha256:de37daa0e4db9b5dccf94408a3422be4f821e380ab70081bd1032cec1e3c91e6", size = 4078640, upload-time = "2026-03-17T21:10:40.275Z" },
-    { url = "https://files.pythonhosted.org/packages/1e/c1/a0295506a521d9942b0f55523781a113e4555420d800a386d5a2eb46a7ad/onyx_devtools-0.7.1-py3-none-win_arm64.whl", hash = "sha256:ab88c53ebda6dff27350316b4ac9bd5f258cd586c2109971a9d976411e1e22ea", size = 3636787, upload-time = "2026-03-17T21:10:37.492Z" },
+    { url = "https://files.pythonhosted.org/packages/22/9e/6957b11555da57d9e97092f4cd8ac09a86666264b0c9491838f4b27db5dc/onyx_devtools-0.7.0-py3-none-macosx_10_12_x86_64.whl", hash = "sha256:ad962a168d46ea11dcde9fa3b37e4f12ec520b4a4cb4d49d8732de110d46c4b6", size = 3998057, upload-time = "2026-03-12T03:09:11.585Z" },
+    { url = "https://files.pythonhosted.org/packages/cd/90/c72f3d06ba677012d77c77de36195b6a32a15c755c79ba0282be74e3c366/onyx_devtools-0.7.0-py3-none-macosx_11_0_arm64.whl", hash = "sha256:e46d252e2b048ff053b03519c3a875998780738d7c334eaa1c9a32ff445e3e1a", size = 3687753, upload-time = "2026-03-12T03:09:11.742Z" },
+    { url = "https://files.pythonhosted.org/packages/10/42/4e9fe36eccf9f76d67ba8f4ff6539196a09cd60351fb63f5865e1544cbfa/onyx_devtools-0.7.0-py3-none-manylinux_2_17_aarch64.whl", hash = "sha256:f280bc9320e1cc310e7d753a371009bfaab02cc0e0cfd78559663b15655b5a50", size = 3560144, upload-time = "2026-03-12T03:12:24.02Z" },
+    { url = "https://files.pythonhosted.org/packages/76/40/36dc12d99760b358c7f39b27361cb18fa9681ffe194107f982d0e1a74016/onyx_devtools-0.7.0-py3-none-manylinux_2_17_x86_64.whl", hash = "sha256:e31df751c7540ae7e70a7fe8e1153c79c31c2254af6aa4c72c0dd54fa381d2ab", size = 3964387, upload-time = "2026-03-12T03:09:11.356Z" },
+    { url = "https://files.pythonhosted.org/packages/34/18/74744230c3820a5a7687335507ca5f1dbebab2c5325805041c1cd5703e6a/onyx_devtools-0.7.0-py3-none-win_amd64.whl", hash = "sha256:541bfd347c2d5b11e7f63ab5001d2594df91d215ad9d07b1562f5e715700f7e6", size = 4068030, upload-time = "2026-03-12T03:09:12.98Z" },
+    { url = "https://files.pythonhosted.org/packages/8c/78/1320436607d3ffcb321ba7b064556c020ea15843a7e7d903fbb7529a71f5/onyx_devtools-0.7.0-py3-none-win_arm64.whl", hash = "sha256:83016330a9d39712431916cc25b2fb2cfcaa0112a55cc4f919d545da3a8974f9", size = 3626409, upload-time = "2026-03-12T03:09:10.222Z" },
 ]
 
 [[package]]
@@ -5458,11 +5458,11 @@ wheels = [
 
 [[package]]
 name = "pyasn1"
-version = "0.6.3"
+version = "0.6.2"
 source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/5c/5f/6583902b6f79b399c9c40674ac384fd9cd77805f9e6205075f828ef11fb2/pyasn1-0.6.3.tar.gz", hash = "sha256:697a8ecd6d98891189184ca1fa05d1bb00e2f84b5977c481452050549c8a72cf", size = 148685, upload-time = "2026-03-17T01:06:53.382Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/fe/b6/6e630dff89739fcd427e3f72b3d905ce0acb85a45d4ec3e2678718a3487f/pyasn1-0.6.2.tar.gz", hash = "sha256:9b59a2b25ba7e4f8197db7686c09fb33e658b98339fadb826e9512629017833b", size = 146586, upload-time = "2026-01-16T18:04:18.534Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/5d/a0/7d793dce3fa811fe047d6ae2431c672364b462850c6235ae306c0efd025f/pyasn1-0.6.3-py3-none-any.whl", hash = "sha256:a80184d120f0864a52a073acc6fc642847d0be408e7c7252f31390c0f4eadcde", size = 83997, upload-time = "2026-03-17T01:06:52.036Z" },
+    { url = "https://files.pythonhosted.org/packages/44/b5/a96872e5184f354da9c84ae119971a0a4c221fe9b27a4d94bd43f2596727/pyasn1-0.6.2-py3-none-any.whl", hash = "sha256:1eb26d860996a18e9b6ed05e7aae0e9fc21619fcee6af91cca9bad4fbea224bf", size = 83371, upload-time = "2026-01-16T18:04:17.174Z" },
 ]
 
 [[package]]

web/lib/opal/src/components/buttons/button/components.tsx

@@ -1,7 +1,7 @@
 import "@opal/components/buttons/button/styles.css";
 import "@opal/components/tooltip.css";
 import { Interactive, type InteractiveStatelessProps } from "@opal/core";
-import type { ContainerSizeVariants, ExtremaSizeVariants } from "@opal/types";
+import type { SizeVariant, WidthVariant } from "@opal/shared";
 import type { TooltipSide } from "@opal/components";
 import type { IconFunctionComponent } from "@opal/types";
 import * as TooltipPrimitive from "@radix-ui/react-tooltip";
@@ -31,7 +31,7 @@ type ButtonProps = InteractiveStatelessProps &
     /**
      * Size preset — controls gap, text size, and Container height/rounding.
      */
-    size?: ContainerSizeVariants;
+    size?: SizeVariant;
 
     /** HTML button type. When provided, Container renders a `<button>` element. */
     type?: "submit" | "button" | "reset";
@@ -40,7 +40,7 @@ type ButtonProps = InteractiveStatelessProps &
     tooltip?: string;
 
     /** Width preset. `"fit"` shrink-wraps, `"full"` stretches to parent width. */
-    width?: ExtremaSizeVariants;
+    width?: WidthVariant;
 
     /** Which side the tooltip appears on. */
     tooltipSide?: TooltipSide;

web/lib/opal/src/components/buttons/icon-wrapper.tsx

@@ -1,4 +1,4 @@
-import type { ContainerSizeVariants } from "@opal/types";
+import type { SizeVariant } from "@opal/shared";
 import type { IconFunctionComponent } from "@opal/types";
 import { cn } from "@opal/utils";
 
@@ -13,7 +13,7 @@ const iconVariants = {
 
 function iconWrapper(
   Icon: IconFunctionComponent | undefined,
-  size: ContainerSizeVariants,
+  size: SizeVariant,
   includeSpacer: boolean
 ) {
   const { padding: p, size: s } = iconVariants[size];

web/lib/opal/src/components/buttons/line-item-button/components.tsx

@@ -6,7 +6,7 @@ import {
   type InteractiveStatefulProps,
   InteractiveContainerRoundingVariant,
 } from "@opal/core";
-import type { ExtremaSizeVariants } from "@opal/types";
+import { type WidthVariant } from "@opal/shared";
 import type { TooltipSide } from "@opal/components";
 import type { DistributiveOmit } from "@opal/types";
 import type { ContentActionProps } from "@opal/layouts/content-action/components";
@@ -51,7 +51,7 @@ type LineItemButtonOwnProps = {
   roundingVariant?: InteractiveContainerRoundingVariant;
 
   /** Container width. @default "full" */
-  width?: ExtremaSizeVariants;
+  width?: WidthVariant;
 
   /** HTML button type. @default "button" */
   type?: "submit" | "button" | "reset";

web/lib/opal/src/components/buttons/open-button/components.tsx

@@ -6,7 +6,7 @@ import {
   type InteractiveStatefulProps,
   type InteractiveStatefulInteraction,
 } from "@opal/core";
-import type { ContainerSizeVariants, ExtremaSizeVariants } from "@opal/types";
+import type { SizeVariant, WidthVariant } from "@opal/shared";
 import type { InteractiveContainerRoundingVariant } from "@opal/core";
 import type { TooltipSide } from "@opal/components";
 import type { IconFunctionComponent, IconProps } from "@opal/types";
@@ -64,10 +64,10 @@ type OpenButtonProps = Omit<InteractiveStatefulProps, "variant"> & {
     /**
      * Size preset — controls gap, text size, and Container height/rounding.
      */
-    size?: ContainerSizeVariants;
+    size?: SizeVariant;
 
     /** Width preset. */
-    width?: ExtremaSizeVariants;
+    width?: WidthVariant;
 
     /**
      * Content justify mode. When `"between"`, icon+label group left and

web/lib/opal/src/components/buttons/select-button/components.tsx

@@ -5,7 +5,7 @@ import {
   useDisabled,
   type InteractiveStatefulProps,
 } from "@opal/core";
-import type { ContainerSizeVariants, ExtremaSizeVariants } from "@opal/types";
+import type { SizeVariant, WidthVariant } from "@opal/shared";
 import type { TooltipSide } from "@opal/components";
 import type { IconFunctionComponent } from "@opal/types";
 import * as TooltipPrimitive from "@radix-ui/react-tooltip";
@@ -48,7 +48,7 @@ type SelectButtonProps = InteractiveStatefulProps &
     /**
      * Size preset — controls gap, text size, and Container height/rounding.
      */
-    size?: ContainerSizeVariants;
+    size?: SizeVariant;
 
     /** HTML button type. Container renders a `<button>` element. */
     type?: "submit" | "button" | "reset";
@@ -57,7 +57,7 @@ type SelectButtonProps = InteractiveStatefulProps &
     tooltip?: string;
 
     /** Width preset. `"fit"` shrink-wraps, `"full"` stretches to parent width. */
-    width?: ExtremaSizeVariants;
+    width?: WidthVariant;
 
     /** Which side the tooltip appears on. */
     tooltipSide?: TooltipSide;

web/lib/opal/src/components/cards/card/components.tsx

@@ -1,6 +1,6 @@
 import "@opal/components/cards/card/styles.css";
-import type { ContainerSizeVariants } from "@opal/types";
-import { containerSizeVariants } from "@opal/shared";
+import type { SizeVariant } from "@opal/shared";
+import { sizeVariants } from "@opal/shared";
 import { cn } from "@opal/utils";
 
 // ---------------------------------------------------------------------------
@@ -26,7 +26,7 @@ type CardProps = {
    *
    * @default "lg"
    */
-  sizeVariant?: ContainerSizeVariants;
+  sizeVariant?: SizeVariant;
 
   /**
    * Background fill intensity.
@@ -59,7 +59,7 @@ type CardProps = {
 // ---------------------------------------------------------------------------
 
 /** Maps a size variant to a rounding class, mirroring the Button pattern. */
-const roundingForSize: Record<ContainerSizeVariants, string> = {
+const roundingForSize: Record<SizeVariant, string> = {
   lg: "rounded-12",
   md: "rounded-08",
   sm: "rounded-08",
@@ -79,7 +79,7 @@ function Card({
   ref,
   children,
 }: CardProps) {
-  const { padding } = containerSizeVariants[sizeVariant];
+  const { padding } = sizeVariants[sizeVariant];
   const rounding = roundingForSize[sizeVariant];
 
   return (

web/lib/opal/src/components/cards/empty-message-card/components.tsx

@@ -1,7 +1,7 @@
 import { Card } from "@opal/components/cards/card/components";
 import { Content } from "@opal/layouts";
 import { SvgEmpty } from "@opal/icons";
-import type { ContainerSizeVariants } from "@opal/types";
+import type { SizeVariant } from "@opal/shared";
 import type { IconFunctionComponent } from "@opal/types";
 
 // ---------------------------------------------------------------------------
@@ -16,7 +16,7 @@ type EmptyMessageCardProps = {
   title: string;
 
   /** Size preset controlling padding and rounding of the card. */
-  sizeVariant?: ContainerSizeVariants;
+  sizeVariant?: SizeVariant;
 
   /** Ref forwarded to the root Card div. */
   ref?: React.Ref<HTMLDivElement>;

web/lib/opal/src/components/index.ts

@@ -52,10 +52,3 @@ export {
   type PaginationProps,
   type PaginationSize,
 } from "@opal/components/pagination/components";
-
-/* Table */
-export { Table } from "@opal/components/table/components";
-
-export { createTableColumns } from "@opal/components/table/columns";
-
-export type { DataTableProps } from "@opal/components/table/types";

web/lib/opal/src/components/pagination/components.tsx

@@ -3,7 +3,7 @@
 import { Button } from "@opal/components";
 import { Disabled } from "@opal/core";
 import { SvgArrowRight, SvgChevronLeft, SvgChevronRight } from "@opal/icons";
-import { containerSizeVariants } from "@opal/shared";
+import { sizeVariants } from "@opal/shared";
 import type { WithoutStyles } from "@opal/types";
 import { cn } from "@opal/utils";
 import * as PopoverPrimitive from "@radix-ui/react-popover";
@@ -252,7 +252,7 @@ function GoToPagePopup({ totalPages, onSubmit, children }: GoToPagePopupProps) {
             autoFocus
             className={cn(
               "w-[7rem] bg-transparent px-1.5 py-1 rounded-08",
-              containerSizeVariants.lg.height,
+              sizeVariants.lg.height,
               "border border-border-02 focus:outline-none focus:border-border-04",
               "font-main-ui-body",
               "text-text-04 placeholder:text-text-02"

web/lib/opal/src/components/table/README.md

@@ -1,82 +0,0 @@
-# Table
-
-Config-driven table component with sorting, pagination, column visibility,
-row selection, drag-and-drop reordering, and server-side mode.
-
-## Usage
-
-```tsx
-import { Table, createTableColumns } from "@opal/components";
-
-interface User {
-  id: string;
-  email: string;
-  name: string | null;
-  status: "active" | "invited";
-}
-
-const tc = createTableColumns<User>();
-
-const columns = [
-  tc.qualifier({ content: "avatar-user", getInitials: (r) => r.name?.[0] ?? "?" }),
-  tc.column("email", {
-    header: "Name",
-    weight: 22,
-    minWidth: 140,
-    cell: (email, row) => <span>{row.name ?? email}</span>,
-  }),
-  tc.column("status", {
-    header: "Status",
-    weight: 14,
-    cell: (status) => <span>{status}</span>,
-  }),
-  tc.actions(),
-];
-
-function UsersTable({ users }: { users: User[] }) {
-  return (
-    <Table
-      data={users}
-      columns={columns}
-      getRowId={(r) => r.id}
-      pageSize={10}
-      footer={{ mode: "summary" }}
-    />
-  );
-}
-```
-
-## Props
-
-| Prop | Type | Default | Description |
-|------|------|---------|-------------|
-| `data` | `TData[]` | required | Row data array |
-| `columns` | `OnyxColumnDef<TData>[]` | required | Column definitions from `createTableColumns()` |
-| `getRowId` | `(row: TData) => string` | required | Unique row identifier |
-| `pageSize` | `number` | `10` | Rows per page (`Infinity` disables pagination) |
-| `size` | `"md" \| "lg"` | `"lg"` | Density variant |
-| `footer` | `DataTableFooterConfig` | — | Footer mode (`"selection"` or `"summary"`) |
-| `initialSorting` | `SortingState` | — | Initial sort state |
-| `initialColumnVisibility` | `VisibilityState` | — | Initial column visibility |
-| `draggable` | `DataTableDraggableConfig` | — | Enable drag-and-drop reordering |
-| `onSelectionChange` | `(ids: string[]) => void` | — | Selection callback |
-| `onRowClick` | `(row: TData) => void` | — | Row click handler |
-| `searchTerm` | `string` | — | Global text filter |
-| `height` | `number \| string` | — | Max scrollable height |
-| `headerBackground` | `string` | — | Sticky header background |
-| `serverSide` | `ServerSideConfig` | — | Server-side pagination/sorting/filtering |
-| `emptyState` | `ReactNode` | — | Empty state content |
-
-## Column Builder
-
-`createTableColumns<TData>()` returns a builder with:
-
-- `tc.qualifier(opts)` — leading avatar/icon/checkbox column
-- `tc.column(accessor, opts)` — data column with sorting/resizing
-- `tc.display(opts)` — non-accessor custom column
-- `tc.actions(opts)` — trailing actions column with visibility/sorting popovers
-
-## Footer Modes
-
-- **`"selection"`** — shows selection count, optional view/clear buttons, count pagination
-- **`"summary"`** — shows "Showing X~Y of Z", list pagination, optional extra element

web/lib/opal/src/components/table/Table.stories.tsx

@@ -1,148 +0,0 @@
-import type { Meta, StoryObj } from "@storybook/react";
-import { Table, createTableColumns } from "@opal/components";
-
-// ---------------------------------------------------------------------------
-// Sample data
-// ---------------------------------------------------------------------------
-
-interface User {
-  id: string;
-  email: string;
-  name: string;
-  role: "admin" | "user" | "viewer";
-  status: "active" | "invited" | "inactive";
-}
-
-const USERS: User[] = [
-  {
-    id: "1",
-    email: "alice@example.com",
-    name: "Alice Johnson",
-    role: "admin",
-    status: "active",
-  },
-  {
-    id: "2",
-    email: "bob@example.com",
-    name: "Bob Smith",
-    role: "user",
-    status: "active",
-  },
-  {
-    id: "3",
-    email: "carol@example.com",
-    name: "Carol White",
-    role: "viewer",
-    status: "invited",
-  },
-  {
-    id: "4",
-    email: "dave@example.com",
-    name: "Dave Brown",
-    role: "user",
-    status: "inactive",
-  },
-  {
-    id: "5",
-    email: "eve@example.com",
-    name: "Eve Davis",
-    role: "admin",
-    status: "active",
-  },
-  {
-    id: "6",
-    email: "frank@example.com",
-    name: "Frank Miller",
-    role: "viewer",
-    status: "active",
-  },
-  {
-    id: "7",
-    email: "grace@example.com",
-    name: "Grace Lee",
-    role: "user",
-    status: "invited",
-  },
-  {
-    id: "8",
-    email: "hank@example.com",
-    name: "Hank Wilson",
-    role: "user",
-    status: "active",
-  },
-  {
-    id: "9",
-    email: "iris@example.com",
-    name: "Iris Taylor",
-    role: "viewer",
-    status: "active",
-  },
-  {
-    id: "10",
-    email: "jack@example.com",
-    name: "Jack Moore",
-    role: "admin",
-    status: "active",
-  },
-  {
-    id: "11",
-    email: "kate@example.com",
-    name: "Kate Anderson",
-    role: "user",
-    status: "inactive",
-  },
-  {
-    id: "12",
-    email: "leo@example.com",
-    name: "Leo Thomas",
-    role: "viewer",
-    status: "active",
-  },
-];
-
-// ---------------------------------------------------------------------------
-// Columns
-// ---------------------------------------------------------------------------
-
-const tc = createTableColumns<User>();
-
-const columns = [
-  tc.qualifier({
-    content: "avatar-user",
-    getInitials: (r) =>
-      r.name
-        .split(" ")
-        .map((n) => n[0])
-        .join(""),
-  }),
-  tc.column("name", { header: "Name", weight: 25, minWidth: 120 }),
-  tc.column("email", { header: "Email", weight: 30, minWidth: 160 }),
-  tc.column("role", { header: "Role", weight: 15, minWidth: 80 }),
-  tc.column("status", { header: "Status", weight: 15, minWidth: 80 }),
-  tc.actions(),
-];
-
-// ---------------------------------------------------------------------------
-// Story
-// ---------------------------------------------------------------------------
-
-const meta: Meta<typeof Table> = {
-  title: "opal/components/Table",
-  component: Table,
-  tags: ["autodocs"],
-};
-
-export default meta;
-type Story = StoryObj<typeof Table>;
-
-export const Default: Story = {
-  render: () => (
-    <Table
-      data={USERS}
-      columns={columns}
-      getRowId={(r) => r.id}
-      pageSize={8}
-      footer={{ mode: "summary" }}
-    />
-  ),
-};

web/lib/opal/src/components/table/TableElement.tsx

@@ -1,70 +0,0 @@
-import { cn } from "@opal/utils";
-import type { WithoutStyles } from "@/types";
-import type { ExtremaSizeVariants, SizeVariants } from "@opal/types";
-
-// ---------------------------------------------------------------------------
-// Types
-// ---------------------------------------------------------------------------
-
-type TableSize = Extract<SizeVariants, "md" | "lg">;
-type TableVariant = "rows" | "card";
-type TableQualifier = null | "icon" | "checkbox";
-type SelectionBehavior = "no-select" | "single-select" | "multi-select";
-
-interface TableProps
-  extends WithoutStyles<React.TableHTMLAttributes<HTMLTableElement>> {
-  ref?: React.Ref<HTMLTableElement>;
-  /** Size preset for the table. @default "lg" */
-  size?: TableSize;
-  /** Visual row variant. @default "card" */
-  variant?: TableVariant;
-  /** Row selection behavior. @default "no-select" */
-  selectionBehavior?: SelectionBehavior;
-  /** Leading qualifier column type. @default null */
-  qualifier?: TableQualifier;
-  /** Height behavior. `"fit"` = shrink to content, `"full"` = fill available space. */
-  heightVariant?: ExtremaSizeVariants;
-  /** Explicit pixel width for the table (e.g. from `table.getTotalSize()`).
-   *  When provided the table uses exactly this width instead of stretching
-   *  to fill its container, which prevents `table-layout: fixed` from
-   *  redistributing extra space across columns on resize. */
-  width?: number;
-}
-
-// ---------------------------------------------------------------------------
-// Component
-// ---------------------------------------------------------------------------
-
-function Table({
-  ref,
-  size = "lg",
-  variant = "card",
-  selectionBehavior = "no-select",
-  qualifier = null,
-  heightVariant,
-  width,
-  ...props
-}: TableProps) {
-  return (
-    <table
-      ref={ref}
-      className={cn("border-separate border-spacing-0", "min-w-full")}
-      style={{ tableLayout: "fixed", width: width ?? undefined }}
-      data-size={size}
-      data-variant={variant}
-      data-selection={selectionBehavior}
-      data-qualifier={qualifier ?? undefined}
-      data-height={heightVariant}
-      {...props}
-    />
-  );
-}
-
-export default Table;
-export type {
-  TableProps,
-  TableSize,
-  TableVariant,
-  TableQualifier,
-  SelectionBehavior,
-};

web/lib/opal/src/core/animations/components.tsx

@@ -1,8 +1,8 @@
 import "@opal/core/animations/styles.css";
 import React, { createContext, useContext, useState, useCallback } from "react";
 import { cn } from "@opal/utils";
-import type { WithoutStyles, ExtremaSizeVariants } from "@opal/types";
-import { widthVariants } from "@opal/shared";
+import type { WithoutStyles } from "@opal/types";
+import { widthVariants, type WidthVariant } from "@opal/shared";
 
 // ---------------------------------------------------------------------------
 // Context-per-group registry
@@ -40,7 +40,7 @@ interface HoverableRootProps
   children: React.ReactNode;
   group: string;
   /** Width preset. @default "auto" */
-  widthVariant?: ExtremaSizeVariants;
+  widthVariant?: WidthVariant;
   /** Ref forwarded to the root `<div>`. */
   ref?: React.Ref<HTMLDivElement>;
 }
@@ -84,7 +84,7 @@ interface HoverableItemProps
 function HoverableRoot({
   group,
   children,
-  widthVariant = "full",
+  widthVariant = "auto",
   ref,
   onMouseEnter: consumerMouseEnter,
   onMouseLeave: consumerMouseLeave,

web/lib/opal/src/core/interactive/container/components.tsx

@@ -5,10 +5,10 @@ import React from "react";
 import { cn } from "@opal/utils";
 import type { WithoutStyles } from "@opal/types";
 import {
-  containerSizeVariants,
-  type ContainerSizeVariants,
+  sizeVariants,
+  type SizeVariant,
   widthVariants,
-  type ExtremaSizeVariants,
+  type WidthVariant,
 } from "@opal/shared";
 import { useDisabled } from "@opal/core/disabled/components";
 
@@ -73,14 +73,14 @@ interface InteractiveContainerProps
    *
    * @default "lg"
    */
-  heightVariant?: ContainerSizeVariants;
+  heightVariant?: SizeVariant;
 
   /**
    * Width preset controlling the container's horizontal size.
    *
    * @default "fit"
    */
-  widthVariant?: ExtremaSizeVariants;
+  widthVariant?: WidthVariant;
 }
 
 // ---------------------------------------------------------------------------
@@ -119,7 +119,7 @@ function InteractiveContainer({
     target?: string;
     rel?: string;
   };
-  const { height, minWidth, padding } = containerSizeVariants[heightVariant];
+  const { height, minWidth, padding } = sizeVariants[heightVariant];
   const sharedProps = {
     ...rest,
     className: cn(

web/lib/opal/src/layouts/content-action/components.tsx

@@ -1,8 +1,5 @@
 import { Content, type ContentProps } from "@opal/layouts/content/components";
-import {
-  containerSizeVariants,
-  type ContainerSizeVariants,
-} from "@opal/shared";
+import { sizeVariants, type SizeVariant } from "@opal/shared";
 import { cn } from "@opal/utils";
 
 // ---------------------------------------------------------------------------
@@ -18,9 +15,9 @@ type ContentActionProps = ContentProps & {
    * Uses the shared `SizeVariant` scale from `@opal/shared`.
    *
    * @default "lg"
-   * @see {@link ContainerSizeVariants} for the full list of presets.
+   * @see {@link SizeVariant} for the full list of presets.
    */
-  paddingVariant?: ContainerSizeVariants;
+  paddingVariant?: SizeVariant;
 };
 
 // ---------------------------------------------------------------------------
@@ -57,7 +54,7 @@ function ContentAction({
   paddingVariant = "lg",
   ...contentProps
 }: ContentActionProps) {
-  const { padding } = containerSizeVariants[paddingVariant];
+  const { padding } = sizeVariants[paddingVariant];
 
   return (
     <div className="flex flex-row items-stretch w-full">

web/lib/opal/src/layouts/content/ContentLg.tsx

@@ -1,7 +1,7 @@
 "use client";
 
 import { Button } from "@opal/components/buttons/button/components";
-import type { ContainerSizeVariants } from "@opal/types";
+import type { SizeVariant } from "@opal/shared";
 import SvgEdit from "@opal/icons/edit";
 import type { IconFunctionComponent } from "@opal/types";
 import { cn } from "@opal/utils";
@@ -25,7 +25,7 @@ interface ContentLgPresetConfig {
   /** Title line-height — also used as icon container min-height (CSS value). */
   lineHeight: string;
   /** Button `size` prop for the edit button. Uses the shared `SizeVariant` scale. */
-  editButtonSize: ContainerSizeVariants;
+  editButtonSize: SizeVariant;
   /** Tailwind padding class for the edit button container. */
   editButtonPadding: string;
 }

web/lib/opal/src/layouts/content/ContentMd.tsx

@@ -2,7 +2,7 @@
 
 import { Button } from "@opal/components/buttons/button/components";
 import { Tag, type TagProps } from "@opal/components/tag/components";
-import type { ContainerSizeVariants } from "@opal/types";
+import type { SizeVariant } from "@opal/shared";
 import SvgAlertCircle from "@opal/icons/alert-circle";
 import SvgAlertTriangle from "@opal/icons/alert-triangle";
 import SvgEdit from "@opal/icons/edit";
@@ -27,7 +27,7 @@ interface ContentMdPresetConfig {
   lineHeight: string;
   gap: string;
   /** Button `size` prop for the edit button. Uses the shared `SizeVariant` scale. */
-  editButtonSize: ContainerSizeVariants;
+  editButtonSize: SizeVariant;
   editButtonPadding: string;
   optionalFont: string;
   /** Aux icon size = lineHeight − 2 × p-0.5. */

web/lib/opal/src/layouts/content/ContentXl.tsx

@@ -1,7 +1,7 @@
 "use client";
 
 import { Button } from "@opal/components/buttons/button/components";
-import type { ContainerSizeVariants } from "@opal/types";
+import type { SizeVariant } from "@opal/shared";
 import SvgEdit from "@opal/icons/edit";
 import type { IconFunctionComponent } from "@opal/types";
 import { cn } from "@opal/utils";
@@ -31,7 +31,7 @@ interface ContentXlPresetConfig {
   /** Title line-height — also used as icon container min-height (CSS value). */
   lineHeight: string;
   /** Button `size` prop for the edit button. Uses the shared `SizeVariant` scale. */
-  editButtonSize: ContainerSizeVariants;
+  editButtonSize: SizeVariant;
   /** Tailwind padding class for the edit button container. */
   editButtonPadding: string;
 }

web/lib/opal/src/layouts/content/components.tsx

@@ -18,8 +18,7 @@ import {
 } from "@opal/layouts/content/ContentMd";
 import type { TagProps } from "@opal/components/tag/components";
 import type { IconFunctionComponent } from "@opal/types";
-import { widthVariants } from "@opal/shared";
-import type { ExtremaSizeVariants } from "@opal/types";
+import { widthVariants, type WidthVariant } from "@opal/shared";
 
 // ---------------------------------------------------------------------------
 // Shared types
@@ -60,7 +59,7 @@ interface ContentBaseProps {
    *
    * @default "fit"
    */
-  widthVariant?: ExtremaSizeVariants;
+  widthVariant?: WidthVariant;
 
   /** When `true`, the title color hooks into `Interactive.Stateful`/`Interactive.Stateless`'s `--interactive-foreground` variable. */
   withInteractive?: boolean;
@@ -129,7 +128,7 @@ function Content(props: ContentProps) {
   const {
     sizePreset = "headline",
     variant = "heading",
-    widthVariant = "full",
+    widthVariant = "auto",
     withInteractive,
     ref,
     ...rest

web/lib/opal/src/shared.ts

@@ -6,12 +6,18 @@
  * circular imports and gives every consumer a single source of truth.
  */
 
-import type {
-  SizeVariants,
-  OverridableExtremaSizeVariants,
-  ContainerSizeVariants,
-  ExtremaSizeVariants,
-} from "@opal/types";
+// ---------------------------------------------------------------------------
+// Size Variants
+//
+// A named scale of size presets (lg → 2xs, plus fit) that map to Tailwind
+// utility classes for height, min-width, and padding.
+//
+// Consumers:
+//   - Interactive.Container  (height + min-width + padding)
+//   - Button                 (icon sizing)
+//   - ContentAction          (padding only)
+//   - Content (ContentXl / ContentLg / ContentMd)  (edit-button size)
+// ---------------------------------------------------------------------------
 
 /**
  * Size-variant scale.
@@ -28,16 +34,7 @@ import type {
  * | `2xs` | 1rem (16px)   | `p-0.5` |
  * | `fit` | h-fit         | `p-0`   |
  */
-type ContainerProperties = {
-  height: string;
-  minWidth: string;
-  padding: string;
-};
-const containerSizeVariants: Record<
-  ContainerSizeVariants,
-  ContainerProperties
-> = {
-  fit: { height: "h-fit", minWidth: "", padding: "p-0" },
+const sizeVariants = {
   lg: { height: "h-[2.25rem]", minWidth: "min-w-[2.25rem]", padding: "p-2" },
   md: { height: "h-[1.75rem]", minWidth: "min-w-[1.75rem]", padding: "p-1" },
   sm: { height: "h-[1.5rem]", minWidth: "min-w-[1.5rem]", padding: "p-1" },
@@ -47,14 +44,18 @@ const containerSizeVariants: Record<
     padding: "p-0.5",
   },
   "2xs": { height: "h-[1rem]", minWidth: "min-w-[1rem]", padding: "p-0.5" },
+  fit: { height: "h-fit", minWidth: "", padding: "p-0" },
 } as const;
 
+/** Named size preset key. */
+type SizeVariant = keyof typeof sizeVariants;
+
 // ---------------------------------------------------------------------------
-// Width/Height Variants
+// Width Variants
 //
-// A named scale of width/height presets that map to Tailwind width/height utility classes.
+// A named scale of width presets that map to Tailwind width utility classes.
 //
-// Consumers (for width):
+// Consumers:
 //   - Interactive.Container  (widthVariant)
 //   - Button                 (width)
 //   - Content                (widthVariant)
@@ -69,31 +70,13 @@ const containerSizeVariants: Record<
  * | `fit`  | `w-fit`        |
  * | `full` | `w-full`       |
  */
-const widthVariants: Record<ExtremaSizeVariants, string> = {
+const widthVariants = {
+  auto: "w-auto",
   fit: "w-fit",
   full: "w-full",
 } as const;
 
-/**
- * Height-variant scale.
- *
- * | Key    | Tailwind class |
- * |--------|----------------|
- * | `auto` | `h-auto`       |
- * | `fit`  | `h-fit`        |
- * | `full` | `h-full`       |
- */
-const heightVariants: Record<ExtremaSizeVariants, string> = {
-  fit: "h-fit",
-  full: "h-full",
-} as const;
+/** Named width preset key. */
+type WidthVariant = keyof typeof widthVariants;
 
-export {
-  type ExtremaSizeVariants,
-  type ContainerSizeVariants,
-  type OverridableExtremaSizeVariants,
-  type SizeVariants,
-  containerSizeVariants,
-  widthVariants,
-  heightVariants,
-};
+export { sizeVariants, type SizeVariant, widthVariants, type WidthVariant };

web/lib/opal/src/types.ts

@@ -1,61 +1,5 @@
 import type { SVGProps } from "react";
 
-// ---------------------------------------------------------------------------
-// Size Variants
-//
-// A named scale of size presets (lg → 2xs, plus fit) that map to Tailwind
-// utility classes for height, min-width, and padding.
-//
-// Consumers:
-//   - Interactive.Container  (height + min-width + padding)
-//   - Button                 (icon sizing)
-//   - ContentAction          (padding only)
-//   - Content (ContentXl / ContentLg / ContentMd)  (edit-button size)
-// ---------------------------------------------------------------------------
-
-// Base Size Types:
-
-/**
- * Full range of size variants.
- *
- * This is the complete scale of size presets available in the design system.
- * Components needing the full range use this type directly.
- */
-export type SizeVariants = "fit" | "full" | "lg" | "md" | "sm" | "xs" | "2xs";
-
-// Convenience Size Types:
-//
-// NOTE (@raunakab + @nmgarza5)
-// There are many components throughout the library that need to "extract" very specific sizings from the full gamut that is available.
-// For those components, we've extracted these below "convenience" types.
-
-/**
- * Size variants for container components (excludes "full").
- *
- * Used by components that control height, min-width, and padding.
- * Excludes "full" since containers need a fixed height preset.
- */
-export type ContainerSizeVariants = Exclude<SizeVariants, "full">;
-
-/**
- * Extreme size variants ("fit" and "full" only).
- *
- * Used for width and height properties that only support extremal values.
- */
-export type ExtremaSizeVariants = Extract<SizeVariants, "fit" | "full">;
-
-/**
- * Size variants with numeric overrides.
- *
- * Allows size specification as a named preset or a custom numeric value.
- * Used in components that need programmatic sizing flexibility.
- */
-export type OverridableExtremaSizeVariants = ExtremaSizeVariants | number;
-
-// ---------------------------------------------------------------------------
-// Icon Props
-// ---------------------------------------------------------------------------
-
 /**
  * Base props for SVG icon components.
  *

web/lib/opal/tsconfig.json

@@ -2,14 +2,7 @@
   "extends": "../../tsconfig.json",
   "compilerOptions": {
     "paths": {
-      "@opal/*": ["./src/*"],
-      // TODO (@raunakab): Remove this once the table component migration is
-      // complete. The table internals still import app-layer modules (e.g.
-      // @/refresh-components/texts/Text, @/refresh-components/Popover) via the
-      // @/ alias. Without this entry the IDE cannot resolve those paths since
-      // opal's tsconfig only defines @opal/*. Once all @/ deps are replaced
-      // with opal-internal equivalents, this line should be deleted.
-      "@/*": ["../../src/*"]
+      "@opal/*": ["./src/*"]
     }
   },
   "include": ["src/**/*"],

web/src/app/craft/onboarding/components/BuildOnboardingModal.tsx

@@ -24,8 +24,10 @@ import {
 } from "@/app/craft/onboarding/constants";
 import { LLMProviderDescriptor } from "@/interfaces/llm";
 import { LLM_PROVIDERS_ADMIN_URL } from "@/lib/llmConfig/constants";
-import { buildOnboardingInitialValues as buildInitialValues } from "@/sections/modals/llmConfig/utils";
-import { testApiKeyHelper } from "@/sections/modals/llmConfig/svc";
+import {
+  buildInitialValues,
+  testApiKeyHelper,
+} from "@/sections/onboarding/components/llmConnectionHelpers";
 import OnboardingInfoPages from "@/app/craft/onboarding/components/OnboardingInfoPages";
 import OnboardingUserInfo from "@/app/craft/onboarding/components/OnboardingUserInfo";
 import OnboardingLlmSetup, {

web/src/app/css/table.css

@@ -5,17 +5,17 @@
 
 /* ---- TableCell ---- */
 
-.tbl-cell[data-size="lg"] {
+.tbl-cell[data-size="regular"] {
   @apply px-1 py-0.5;
 }
-.tbl-cell[data-size="md"] {
+.tbl-cell[data-size="small"] {
   @apply pl-0.5 pr-1.5 py-1.5;
 }
 
-.tbl-cell-inner[data-size="lg"] {
+.tbl-cell-inner[data-size="regular"] {
   @apply h-10 px-1;
 }
-.tbl-cell-inner[data-size="md"] {
+.tbl-cell-inner[data-size="small"] {
   @apply h-6 px-0.5;
 }
 
@@ -25,10 +25,10 @@
   @apply relative sticky top-0 z-20;
   background: var(--table-header-bg, transparent);
 }
-.table-head[data-size="lg"] {
+.table-head[data-size="regular"] {
   @apply px-2 py-1;
 }
-.table-head[data-size="md"] {
+.table-head[data-size="small"] {
   @apply p-1.5;
 }
 .table-head[data-bottom-border] {
@@ -36,15 +36,15 @@
 }
 
 /* Inner text wrapper */
-.table-head[data-size="lg"] .table-head-label {
+.table-head[data-size="regular"] .table-head-label {
   @apply py-2 px-0.5;
 }
-.table-head[data-size="md"] .table-head-label {
+.table-head[data-size="small"] .table-head-label {
   @apply py-1;
 }
 
 /* Sort button wrapper */
-.table-head[data-size="lg"] .table-head-sort {
+.table-head[data-size="regular"] .table-head-sort {
   @apply py-1.5;
 }
 
@@ -112,14 +112,14 @@
   @apply w-px whitespace-nowrap py-1 sticky top-0 z-20;
   background: var(--table-header-bg, transparent);
 }
-.tbl-qualifier[data-type="head"][data-size="md"] {
+.tbl-qualifier[data-type="head"][data-size="small"] {
   @apply py-0.5;
 }
 
 .tbl-qualifier[data-type="cell"] {
   @apply w-px whitespace-nowrap py-1 pl-1;
 }
-.tbl-qualifier[data-type="cell"][data-size="md"] {
+.tbl-qualifier[data-type="cell"][data-size="small"] {
   @apply py-0.5 pl-0.5;
 }
 
@@ -135,9 +135,9 @@
 
 /* ---- Footer ---- */
 
-.table-footer[data-size="lg"] {
+.table-footer[data-size="regular"] {
   @apply min-h-[2.75rem];
 }
-.table-footer[data-size="md"] {
+.table-footer[data-size="small"] {
   @apply min-h-[2.25rem];
 }

web/src/app/globals.css

@@ -16,6 +16,7 @@
 @import "css/sizes.css";
 @import "css/square-button.css";
 @import "css/switch.css";
+@import "css/table.css";
 @import "css/z-index.css";
 
 /* KH Teka Font */

web/src/components/admin/connectors/ConnectorTitle.tsx

@@ -14,7 +14,7 @@ import Link from "next/link";
 interface ConnectorTitleProps {
   connector: Connector<any>;
   ccPairId: number;
-  ccPairName: string;
+  ccPairName: string | null | undefined;
   isPublic?: boolean;
   owner?: string;
   isLink?: boolean;

web/src/hooks/useLLMProviders.ts

@@ -126,38 +126,6 @@ export function useAdminLLMProviders() {
  * - `error` — The SWR error object, if any.
  * - `mutate` — SWR `mutate` function to trigger a revalidation.
  */
-/**
- * Fetches the descriptor for a single well-known (built-in) LLM provider.
- *
- * Hits `GET /api/admin/llm/built-in/options/{providerEndpoint}` which returns
- * the provider descriptor including its known models and the recommended
- * default model.
- *
- * Used inside individual provider modals to pre-populate model lists
- * before the user has entered credentials.
- *
- * @param providerEndpoint - The provider's API endpoint name (e.g. "openai", "anthropic").
- *   Pass `null` to suppress the request.
- */
-export function useWellKnownLLMProvider(providerEndpoint: string | null) {
-  const { data, error, isLoading } = useSWR<WellKnownLLMProviderDescriptor>(
-    providerEndpoint
-      ? `/api/admin/llm/built-in/options/${providerEndpoint}`
-      : null,
-    errorHandlingFetcher,
-    {
-      revalidateOnFocus: false,
-      dedupingInterval: 60000,
-    }
-  );
-
-  return {
-    wellKnownLLMProvider: data ?? null,
-    isLoading,
-    error,
-  };
-}
-
 export function useWellKnownLLMProviders() {
   const {
     data: wellKnownLLMProviders,

web/src/hooks/useVoiceRecorder.ts

@@ -1,6 +1,6 @@
 import { useState, useRef, useCallback, useEffect } from "react";
 
-import { INTERNAL_URL, IS_DEV } from "@/lib/constants";
+import { IS_DEV } from "@/lib/constants";
 
 // Target format for OpenAI Realtime API
 const TARGET_SAMPLE_RATE = 24000;
@@ -247,7 +247,7 @@ class VoiceRecorderSession {
     const { token } = await tokenResponse.json();
 
     const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
-    const host = IS_DEV ? new URL(INTERNAL_URL).host : window.location.host;
+    const host = IS_DEV ? "localhost:8080" : window.location.host;
     const path = IS_DEV
       ? "/voice/transcribe/stream"
       : "/api/voice/transcribe/stream";

web/src/interfaces/llm.ts

@@ -1,8 +1,3 @@
-import type {
-  OnboardingState,
-  OnboardingActions,
-} from "@/interfaces/onboarding";
-
 export enum LLMProviderName {
   OPENAI = "openai",
   ANTHROPIC = "anthropic",
@@ -114,19 +109,11 @@ export interface LLMProviderResponse<T> {
   default_vision: DefaultModel | null;
 }
 
-export type LLMModalVariant = "onboarding" | "llm-configuration";
-
 export interface LLMProviderFormProps {
-  variant?: LLMModalVariant;
   existingLlmProvider?: LLMProviderView;
   shouldMarkAsDefault?: boolean;
   open?: boolean;
   onOpenChange?: (open: boolean) => void;
-
-  // Onboarding-specific (only when variant === "onboarding")
-  onboardingState?: OnboardingState;
-  onboardingActions?: OnboardingActions;
-  llmDescriptor?: WellKnownLLMProviderDescriptor;
 }
 
 // Param types for model fetching functions - use snake_case to match API structure
@@ -144,6 +131,14 @@ export interface OllamaFetchParams {
   signal?: AbortSignal;
 }
 
+export interface LMStudioFetchParams {
+  api_base?: string;
+  api_key?: string;
+  api_key_changed?: boolean;
+  provider_name?: string;
+  signal?: AbortSignal;
+}
+
 export interface OpenRouterFetchParams {
   api_base?: string;
   api_key?: string;
@@ -166,18 +161,10 @@ export interface VertexAIFetchParams {
   model_configurations?: ModelConfiguration[];
 }
 
-export interface LMStudioFetchParams {
-  api_base?: string;
-  api_key?: string;
-  api_key_changed?: boolean;
-  provider_name?: string;
-  signal?: AbortSignal;
-}
-
 export type FetchModelsParams =
   | BedrockFetchParams
   | OllamaFetchParams
+  | LMStudioFetchParams
   | OpenRouterFetchParams
   | LiteLLMProxyFetchParams
-  | VertexAIFetchParams
-  | LMStudioFetchParams;
+  | VertexAIFetchParams;

web/src/lib/credential.ts

@@ -87,7 +87,7 @@ export async function forceDeleteCredential<T>(credentialId: number) {
 export function linkCredential(
   connectorId: number,
   credentialId: number,
-  name: string,
+  name?: string,
   accessType?: AccessType,
   groups?: number[],
   autoSyncOptions?: Record<string, any>,
@@ -101,7 +101,7 @@ export function linkCredential(
         "Content-Type": "application/json",
       },
       body: JSON.stringify({
-        name,
+        name: name || null,
         access_type: accessType !== undefined ? accessType : "public",
         groups: groups || null,
         auto_sync_options: autoSyncOptions || null,

web/src/lib/streamingTTS.ts

@@ -3,7 +3,7 @@
  * Plays audio chunks as they arrive for smooth, low-latency playback.
  */
 
-import { INTERNAL_URL, IS_DEV } from "@/lib/constants";
+import { IS_DEV } from "@/lib/constants";
 
 /**
  * HTTPStreamingTTSPlayer - Uses HTTP streaming with MediaSource Extensions
@@ -384,7 +384,7 @@ export class WebSocketStreamingTTSPlayer {
     const { token } = await tokenResponse.json();
 
     const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
-    const host = IS_DEV ? new URL(INTERNAL_URL).host : window.location.host;
+    const host = IS_DEV ? "localhost:8080" : window.location.host;
     const path = IS_DEV
       ? "/voice/synthesize/stream"
       : "/api/voice/synthesize/stream";

web/src/lib/types.ts

@@ -184,7 +184,7 @@ export interface DocumentBoostStatus {
 
 export interface FailedConnectorIndexingStatus {
   cc_pair_id: number;
-  name: string;
+  name: string | null;
   error_msg: string | null;
   is_deletable: boolean;
   connector_id: number;
@@ -207,7 +207,7 @@ export interface IndexAttemptSnapshot {
 
 export interface ConnectorStatus<ConnectorConfigType, ConnectorCredentialType> {
   cc_pair_id: number;
-  name: string;
+  name: string | null;
   connector: Connector<ConnectorConfigType>;
   credential: Credential<ConnectorCredentialType>;
   access_type: AccessType;
@@ -230,7 +230,7 @@ export interface ConnectorIndexingStatus<
 
 export interface ConnectorIndexingStatusLite {
   cc_pair_id: number;
-  name: string;
+  name: string | null;
   source: ValidSources;
   access_type: AccessType;
   in_progress: boolean;
@@ -343,7 +343,7 @@ export interface DeletionAttemptSnapshot {
 // DOCUMENT SETS
 export interface CCPairDescriptor<ConnectorType, CredentialType> {
   id: number;
-  name: string;
+  name: string | null;
   connector: Connector<ConnectorType>;
   credential: Credential<CredentialType>;
   access_type: AccessType;
@@ -364,7 +364,7 @@ export interface FederatedConnectorDescriptor {
 // Simplified interfaces with minimal data
 export interface CCPairSummary {
   id: number;
-  name: string;
+  name: string | null;
   source: ValidSources;
   access_type: AccessType;
 }

web/src/providers/VoiceModeProvider.tsx

@@ -10,10 +10,15 @@ import React, {
 } from "react";
 import { useUser } from "@/providers/UserProvider";
 import { useVoiceStatus } from "@/hooks/useVoiceStatus";
-import { INTERNAL_URL, IS_DEV } from "@/lib/constants";
 
 // --- TTS Configuration Constants ---
 
+/** Dev server port - used to detect Next.js dev environment */
+const DEV_PORT = "3000";
+
+/** Backend port for direct WebSocket connection in development */
+const BACKEND_PORT = "8080";
+
 /** WebSocket path for TTS streaming (backend-direct, used in dev) */
 const TTS_WS_PATH = "/voice/synthesize/stream";
 
@@ -466,8 +471,9 @@ export function VoiceModeProvider({ children }: { children: React.ReactNode }) {
     // WebSocket connections, so we connect directly to the backend (port 8080).
     // In production, the reverse proxy handles the /api prefix routing.
     const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
-    const host = IS_DEV ? new URL(INTERNAL_URL).host : window.location.host;
-    const path = IS_DEV ? TTS_WS_PATH : TTS_WS_PATH_PROXIED;
+    const isDev = window.location.port === DEV_PORT;
+    const host = isDev ? "localhost:" + BACKEND_PORT : window.location.host;
+    const path = isDev ? TTS_WS_PATH : TTS_WS_PATH_PROXIED;
     // Auth: the token query param is validated server-side by
     // current_user_from_websocket (single-use, 60s TTL, same checks as HTTP auth).
     return `${protocol}//${host}${path}?token=${encodeURIComponent(token)}`;

web/src/refresh-components/form/FormikFields.stories.tsx

@@ -209,7 +209,12 @@ export const PasswordInput: Story = {
   name: "PasswordInputTypeInField",
   render: () => (
     <FormikWrapper initialValues={{ apiKey: "" }}>
-      <PasswordInputTypeInField name="apiKey" placeholder="sk-..." />
+      <PasswordInputTypeInField
+        name="apiKey"
+        label="API Key"
+        subtext="Your key is stored encrypted."
+        placeholder="sk-..."
+      />
     </FormikWrapper>
   ),
 };

web/src/refresh-components/form/PasswordInputTypeInField.tsx

@@ -5,25 +5,34 @@ import PasswordInputTypeIn, {
   PasswordInputTypeInProps,
 } from "@/refresh-components/inputs/PasswordInputTypeIn";
 import { useOnChangeEvent, useOnBlurEvent } from "@/hooks/formHooks";
+import { FieldLabel } from "@/components/Field";
 
 export interface PasswordInputTypeInFieldProps
   extends Omit<PasswordInputTypeInProps, "value"> {
   name: string;
+  /** Optional label to display above the input */
+  label?: string;
+  /** Optional subtext to display below the label */
+  subtext?: string;
 }
 
 export default function PasswordInputTypeInField({
   name,
+  label,
+  subtext,
   onChange: onChangeProp,
   onBlur: onBlurProp,
+  placeholder,
   ...inputProps
 }: PasswordInputTypeInFieldProps) {
   const [field, meta] = useField(name);
   const onChange = useOnChangeEvent(name, onChangeProp);
   const onBlur = useOnBlurEvent(name, onBlurProp);
   const hasError = meta.touched && meta.error;
+  // Don't show error styling for disabled fields
   const showError = hasError && !inputProps.disabled;
 
-  return (
+  const input = (
     <PasswordInputTypeIn
       {...inputProps}
       id={name}
@@ -31,7 +40,19 @@ export default function PasswordInputTypeInField({
       value={field.value ?? ""}
       onChange={onChange}
       onBlur={onBlur}
+      placeholder={placeholder ?? label ?? "API Key"}
       error={showError ? true : inputProps.error}
     />
   );
+
+  if (!label) {
+    return input;
+  }
+
+  return (
+    <div className="w-full flex flex-col gap-1">
+      <FieldLabel name={name} label={label} subtext={subtext} />
+      {input}
+    </div>
+  );
 }

web/src/refresh-components/table/ActionsContainer.tsx

@@ -1,5 +1,5 @@
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 
 interface ActionsContainerProps {
   type: "head" | "cell";

web/src/refresh-components/table/ColumnVisibilityPopover.tsx

@@ -20,13 +20,13 @@ import Divider from "@/refresh-components/Divider";
 interface ColumnVisibilityPopoverProps<TData extends RowData = RowData> {
   table: Table<TData>;
   columnVisibility: VisibilityState;
-  size?: "md" | "lg";
+  size?: "regular" | "small";
 }
 
 function ColumnVisibilityPopover<TData extends RowData>({
   table,
   columnVisibility,
-  size = "lg",
+  size = "regular",
 }: ColumnVisibilityPopoverProps<TData>) {
   const [open, setOpen] = useState(false);
   const hideableColumns = table
@@ -39,8 +39,8 @@ function ColumnVisibilityPopover<TData extends RowData>({
         <Button
           icon={SvgColumn}
           interaction={open ? "hover" : "rest"}
-          size={size === "md" ? "sm" : "md"}
-          prominence="tertiary"
+          size={size === "small" ? "sm" : "md"}
+          prominence="internal"
           tooltip="Columns"
         />
       </Popover.Trigger>
@@ -80,7 +80,7 @@ function ColumnVisibilityPopover<TData extends RowData>({
 // ---------------------------------------------------------------------------
 
 interface CreateColumnVisibilityColumnOptions {
-  size?: "md" | "lg";
+  size?: "regular" | "small";
 }
 
 function createColumnVisibilityColumn<TData>(

web/src/refresh-components/table/DataTable.tsx

@@ -1,30 +1,30 @@
 "use client";
 "use no memo";
 
-import "@opal/components/table/styles.css";
-
 import { useEffect, useMemo } from "react";
 import { flexRender } from "@tanstack/react-table";
-import useDataTable, { toOnyxSortDirection } from "./hooks/useDataTable";
-import useColumnWidths from "./hooks/useColumnWidths";
-import useDraggableRows from "./hooks/useDraggableRows";
-import TableElement from "./TableElement";
-import TableHeader from "./TableHeader";
-import TableBody from "./TableBody";
-import TableRow from "./TableRow";
-import TableHead from "./TableHead";
-import TableCell from "./TableCell";
-import TableQualifier from "./TableQualifier";
-import QualifierContainer from "./QualifierContainer";
-import ActionsContainer from "./ActionsContainer";
-import DragOverlayRow from "./DragOverlayRow";
-import Footer from "./Footer";
-import { TableSizeProvider } from "./TableSizeContext";
-import { ColumnVisibilityPopover } from "./ColumnVisibilityPopover";
-import { SortingPopover } from "./SortingPopover";
-import type { WidthConfig } from "./hooks/useColumnWidths";
+import useDataTable, {
+  toOnyxSortDirection,
+} from "@/refresh-components/table/hooks/useDataTable";
+import useColumnWidths from "@/refresh-components/table/hooks/useColumnWidths";
+import useDraggableRows from "@/refresh-components/table/hooks/useDraggableRows";
+import Table from "@/refresh-components/table/Table";
+import TableHeader from "@/refresh-components/table/TableHeader";
+import TableBody from "@/refresh-components/table/TableBody";
+import TableRow from "@/refresh-components/table/TableRow";
+import TableHead from "@/refresh-components/table/TableHead";
+import TableCell from "@/refresh-components/table/TableCell";
+import TableQualifier from "@/refresh-components/table/TableQualifier";
+import QualifierContainer from "@/refresh-components/table/QualifierContainer";
+import ActionsContainer from "@/refresh-components/table/ActionsContainer";
+import DragOverlayRow from "@/refresh-components/table/DragOverlayRow";
+import Footer from "@/refresh-components/table/Footer";
+import { TableSizeProvider } from "@/refresh-components/table/TableSizeContext";
+import { ColumnVisibilityPopover } from "@/refresh-components/table/ColumnVisibilityPopover";
+import { SortingPopover } from "@/refresh-components/table/SortingPopover";
+import type { WidthConfig } from "@/refresh-components/table/hooks/useColumnWidths";
 import type { ColumnDef } from "@tanstack/react-table";
-import { cn } from "@opal/utils";
+import { cn } from "@/lib/utils";
 import type {
   DataTableProps,
   DataTableFooterConfig,
@@ -32,8 +32,8 @@ import type {
   OnyxDataColumn,
   OnyxQualifierColumn,
   OnyxActionsColumn,
-} from "./types";
-import type { TableSize } from "./TableSizeContext";
+} from "@/refresh-components/table/types";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 
 // ---------------------------------------------------------------------------
 // Internal: resolve size-dependent widths and build TanStack columns
@@ -116,7 +116,7 @@ function processColumns<TData>(
  * <DataTable data={items} columns={columns} footer={{ mode: "selection" }} />
  * ```
  */
-export function Table<TData>(props: DataTableProps<TData>) {
+export default function DataTable<TData>(props: DataTableProps<TData>) {
   const {
     data,
     columns,
@@ -126,7 +126,7 @@ export function Table<TData>(props: DataTableProps<TData>) {
     initialColumnVisibility,
     draggable,
     footer,
-    size = "lg",
+    size = "regular",
     onSelectionChange,
     onRowClick,
     searchTerm,
@@ -303,7 +303,7 @@ export function Table<TData>(props: DataTableProps<TData>) {
               : undefined),
           }}
         >
-          <TableElement
+          <Table
             width={
               Object.keys(columnWidths).length > 0
                 ? Object.values(columnWidths).reduce((sum, w) => sum + w, 0)
@@ -539,7 +539,7 @@ export function Table<TData>(props: DataTableProps<TData>) {
                 );
               })}
             </TableBody>
-          </TableElement>
+          </Table>
         </div>
 
         {footer && renderFooter(footer)}

web/src/refresh-components/table/DragOverlayRow.tsx

@@ -1,11 +1,14 @@
 import { memo } from "react";
 import { type Row, flexRender } from "@tanstack/react-table";
-import TableRow from "./TableRow";
-import TableCell from "./TableCell";
-import QualifierContainer from "./QualifierContainer";
-import TableQualifier from "./TableQualifier";
-import ActionsContainer from "./ActionsContainer";
-import type { OnyxColumnDef, OnyxQualifierColumn } from "./types";
+import TableRow from "@/refresh-components/table/TableRow";
+import TableCell from "@/refresh-components/table/TableCell";
+import QualifierContainer from "@/refresh-components/table/QualifierContainer";
+import TableQualifier from "@/refresh-components/table/TableQualifier";
+import ActionsContainer from "@/refresh-components/table/ActionsContainer";
+import type {
+  OnyxColumnDef,
+  OnyxQualifierColumn,
+} from "@/refresh-components/table/types";
 
 interface DragOverlayRowProps<TData> {
   row: Row<TData>;

web/src/refresh-components/table/Footer.tsx

@@ -1,10 +1,9 @@
 "use client";
 
-import { cn } from "@opal/utils";
+import { cn } from "@/lib/utils";
 import { Button, Pagination } from "@opal/components";
 import Text from "@/refresh-components/texts/Text";
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
 import { SvgEye, SvgXCircle } from "@opal/icons";
 import type { ReactNode } from "react";
 
@@ -41,8 +40,6 @@ interface FooterSelectionModeProps {
   totalPages: number;
   /** Called when the user navigates to a different page. */
   onPageChange: (page: number) => void;
-  /** Controls overall footer sizing. `"lg"` (default) or `"md"`. */
-  size?: TableSize;
   className?: string;
 }
 
@@ -100,7 +97,7 @@ function getSelectionMessage(
  */
 export default function Footer(props: FooterProps) {
   const resolvedSize = useTableSize();
-  const isSmall = resolvedSize === "md";
+  const isSmall = resolvedSize === "small";
   return (
     <div
       className={cn(
@@ -144,8 +141,8 @@ export default function Footer(props: FooterProps) {
             currentPage={props.currentPage}
             totalPages={props.totalPages}
             onChange={props.onPageChange}
-            units="items"
-            size={isSmall ? "sm" : "md"}
+            units="item(s)"
+            size={isSmall ? "md" : "lg"}
           />
         ) : (
           <Pagination

web/src/refresh-components/table/QualifierContainer.tsx

@@ -1,5 +1,5 @@
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 
 interface QualifierContainerProps {
   type: "head" | "cell";

web/src/refresh-components/table/README.md

@@ -0,0 +1,462 @@
+# DataTable
+
+Config-driven table built on [TanStack Table](https://tanstack.com/table). Handles column sizing (weight-based proportional distribution), drag-and-drop row reordering, pagination, row selection, column visibility, and sorting out of the box.
+
+## Quick Start
+
+```tsx
+import DataTable from "@/refresh-components/table/DataTable";
+import { createTableColumns } from "@/refresh-components/table/columns";
+
+interface Person {
+  name: string;
+  email: string;
+  role: string;
+}
+
+// Define columns at module scope (stable reference, no re-renders)
+const tc = createTableColumns<Person>();
+const columns = [
+  tc.qualifier(),
+  tc.column("name", { header: "Name", weight: 30, minWidth: 120 }),
+  tc.column("email", { header: "Email", weight: 40, minWidth: 150 }),
+  tc.column("role", { header: "Role", weight: 30, minWidth: 80 }),
+  tc.actions(),
+];
+
+function PeopleTable({ data }: { data: Person[] }) {
+  return (
+    <DataTable
+      data={data}
+      columns={columns}
+      pageSize={10}
+      footer={{ mode: "selection" }}
+    />
+  );
+}
+```
+
+## Column Builder API
+
+`createTableColumns<TData>()` returns a typed builder with four methods. Each returns an `OnyxColumnDef<TData>` that you pass to the `columns` prop.
+
+### `tc.qualifier(config?)`
+
+Leading column for avatars, icons, images, or checkboxes.
+
+| Option | Type | Default | Description |
+|---|---|---|---|
+| `content` | `"simple" \| "icon" \| "image" \| "avatar-icon" \| "avatar-user"` | `"simple"` | Body row content type |
+| `headerContentType` | same as `content` | `"simple"` | Header row content type |
+| `getInitials` | `(row: TData) => string` | - | Extract initials (for `"avatar-user"`) |
+| `getIcon` | `(row: TData) => IconFunctionComponent` | - | Extract icon (for `"icon"` / `"avatar-icon"`) |
+| `getImageSrc` | `(row: TData) => string` | - | Extract image src (for `"image"`) |
+| `selectable` | `boolean` | `true` | Show selection checkboxes |
+| `header` | `boolean` | `true` | Render qualifier content in the header |
+
+Width is fixed: 56px at `"regular"` size, 40px at `"small"`.
+
+```ts
+tc.qualifier({
+  content: "avatar-user",
+  getInitials: (row) => row.initials,
+})
+```
+
+### `tc.column(accessor, config)`
+
+Data column with sorting, resizing, and hiding. The `accessor` is a type-safe deep key into `TData`.
+
+| Option | Type | Default | Description |
+|---|---|---|---|
+| `header` | `string` | **required** | Column header label |
+| `cell` | `(value: TValue, row: TData) => ReactNode` | renders value as string | Custom cell renderer |
+| `enableSorting` | `boolean` | `true` | Allow sorting |
+| `enableResizing` | `boolean` | `true` | Allow column resize |
+| `enableHiding` | `boolean` | `true` | Allow hiding via actions popover |
+| `icon` | `(sorted: SortDirection) => IconFunctionComponent` | - | Override the sort indicator icon |
+| `weight` | `number` | `20` | Proportional width weight |
+| `minWidth` | `number` | `50` | Minimum width in pixels |
+
+```ts
+tc.column("email", {
+  header: "Email",
+  weight: 28,
+  minWidth: 150,
+  cell: (value) => <Content sizePreset="main-ui" variant="body" title={value} prominence="muted" />,
+})
+```
+
+### `tc.displayColumn(config)`
+
+Non-accessor column for custom content (e.g. computed values, action buttons per row).
+
+| Option | Type | Default | Description |
+|---|---|---|---|
+| `id` | `string` | **required** | Unique column ID |
+| `header` | `string` | - | Optional header label |
+| `cell` | `(row: TData) => ReactNode` | **required** | Cell renderer |
+| `width` | `ColumnWidth` | **required** | `{ weight, minWidth? }` or `{ fixed }` |
+| `enableHiding` | `boolean` | `true` | Allow hiding |
+
+```ts
+tc.displayColumn({
+  id: "fullName",
+  header: "Full Name",
+  cell: (row) => `${row.firstName} ${row.lastName}`,
+  width: { weight: 25, minWidth: 100 },
+})
+```
+
+### `tc.actions(config?)`
+
+Fixed-width column rendered at the trailing edge. Houses column visibility and sorting popovers in the header.
+
+| Option | Type | Default | Description |
+|---|---|---|---|
+| `showColumnVisibility` | `boolean` | `true` | Show the column visibility popover |
+| `showSorting` | `boolean` | `true` | Show the sorting popover |
+| `sortingFooterText` | `string` | - | Footer text inside the sorting popover |
+| `cell` | `(row: TData) => ReactNode` | - | Row-level cell renderer for action buttons |
+
+Width is fixed: 88px at `"regular"`, 20px at `"small"`.
+
+```ts
+tc.actions({
+  sortingFooterText: "Everyone will see agents in this order.",
+})
+```
+
+Row-level actions — the `cell` callback receives the row data and renders content in each body row. Clicks inside the cell automatically call `stopPropagation`, so they won't trigger row selection.
+
+```tsx
+tc.actions({
+  cell: (row) => (
+    <div className="flex gap-x-1">
+      <IconButton icon={SvgPencil} onClick={() => openEdit(row.id)} />
+      <IconButton icon={SvgTrash} onClick={() => confirmDelete(row.id)} />
+    </div>
+  ),
+})
+```
+
+## DataTable Props
+
+`DataTableProps<TData>`:
+
+| Prop | Type | Default | Description |
+|---|---|---|---|
+| `data` | `TData[]` | **required** | Row data |
+| `columns` | `OnyxColumnDef<TData>[]` | **required** | Columns from `createTableColumns()` |
+| `pageSize` | `number` | `10` (with footer) or `data.length` (without) | Rows per page. `Infinity` disables pagination |
+| `initialSorting` | `SortingState` | `[]` | TanStack sorting state |
+| `initialColumnVisibility` | `VisibilityState` | `{}` | Map of column ID to `false` to hide initially |
+| `draggable` | `DataTableDraggableConfig<TData>` | - | Enable drag-and-drop (see below) |
+| `footer` | `DataTableFooterConfig` | - | Footer mode (see below) |
+| `size` | `"regular" \| "small"` | `"regular"` | Table density variant |
+| `onRowClick` | `(row: TData) => void` | toggles selection | Called on row click, replaces default selection toggle |
+| `height` | `number \| string` | - | Max height for scrollable body (header stays pinned). `300` or `"50vh"` |
+| `headerBackground` | `string` | - | CSS color for the sticky header (prevents content showing through) |
+| `searchTerm` | `string` | - | Search term for client-side global text filtering (case-insensitive match across all accessor columns) |
+| `serverSide` | `ServerSideConfig` | - | Enable server-side mode for manual pagination, sorting, and filtering ([see below](#server-side-mode)) |
+
+## Footer Config
+
+The `footer` prop accepts a discriminated union on `mode`.
+
+### Selection mode
+
+For tables with selectable rows. Shows a selection message + count pagination.
+
+```ts
+footer={{
+  mode: "selection",
+  multiSelect: true,        // default true
+  onView: () => { ... },    // optional "View" button
+  onClear: () => { ... },   // optional "Clear" button (falls back to default clearSelection)
+}}
+```
+
+### Summary mode
+
+For read-only tables. Shows "Showing X~Y of Z" + list pagination.
+
+```ts
+footer={{ mode: "summary" }}
+```
+
+## Draggable Config
+
+Enable drag-and-drop row reordering. DnD is automatically disabled when column sorting is active.
+
+```ts
+<DataTable
+  data={items}
+  columns={columns}
+  draggable={{
+    getRowId: (row) => row.id,
+    onReorder: (ids, changedOrders) => {
+      // ids: new ordered array of all row IDs
+      // changedOrders: { [id]: newIndex } for rows that moved
+      setItems(ids.map((id) => items.find((r) => r.id === id)!));
+    },
+  }}
+/>
+```
+
+| Option | Type | Description |
+|---|---|---|
+| `getRowId` | `(row: TData) => string` | Extract a unique string ID from each row |
+| `onReorder` | `(ids: string[], changedOrders: Record<string, number>) => void \| Promise<void>` | Called after a successful reorder |
+
+## Server-Side Mode
+
+Pass the `serverSide` prop to switch from client-side to server-side pagination, sorting, and filtering. In this mode `data` should contain **only the current page slice** — TanStack operates with `manualPagination`, `manualSorting`, and `manualFiltering` enabled. Drag-and-drop is automatically disabled.
+
+### `ServerSideConfig`
+
+| Prop | Type | Description |
+|---|---|---|
+| `totalItems` | `number` | Total row count from the server, used to compute page count |
+| `isLoading` | `boolean` | Shows a loading overlay (opacity + pointer-events-none) while data is being fetched |
+| `onSortingChange` | `(sorting: SortingState) => void` | Fired when the user clicks a column header |
+| `onPaginationChange` | `(pageIndex: number, pageSize: number) => void` | Fired on page navigation and on automatic resets from sort/search changes |
+| `onSearchTermChange` | `(searchTerm: string) => void` | Fired when the `searchTerm` prop changes |
+
+### Callback contract
+
+The callbacks fire in a predictable order:
+
+- **Sort change** — `onSortingChange` fires first, then the page resets to 0 and `onPaginationChange(0, pageSize)` fires.
+- **Page navigation** — only `onPaginationChange` fires.
+- **Search change** — `onSearchTermChange` fires, and the page resets to 0. `onPaginationChange` only fires if the page was actually on a non-zero page. When already on page 0, `searchTerm` drives the re-fetch independently (e.g. via your SWR key) — no `onPaginationChange` is needed.
+
+Your data-fetching layer should include `searchTerm` in its fetch dependencies (e.g. SWR key) so that search changes trigger re-fetches regardless of pagination state.
+
+### Full example
+
+```tsx
+import { useState } from "react";
+import useSWR from "swr";
+import type { SortingState } from "@tanstack/react-table";
+import DataTable from "@/refresh-components/table/DataTable";
+import { createTableColumns } from "@/refresh-components/table/columns";
+import InputTypeIn from "@/refresh-components/inputs/InputTypeIn";
+
+interface User {
+  id: string;
+  name: string;
+  email: string;
+}
+
+const tc = createTableColumns<User>();
+const columns = [
+  tc.qualifier(),
+  tc.column("name", { header: "Name", weight: 40, minWidth: 120 }),
+  tc.column("email", { header: "Email", weight: 60, minWidth: 150 }),
+  tc.actions(),
+];
+
+function UsersTable() {
+  const [searchTerm, setSearchTerm] = useState("");
+  const [sorting, setSorting] = useState<SortingState>([]);
+  const [pageIndex, setPageIndex] = useState(0);
+  const [pageSize, setPageSize] = useState(10);
+
+  const { data: response, isLoading } = useSWR(
+    ["/api/users", sorting, pageIndex, pageSize, searchTerm],
+    ([url, sorting, pageIndex, pageSize, searchTerm]) =>
+      fetch(
+        `${url}?` +
+          new URLSearchParams({
+            page: String(pageIndex),
+            size: String(pageSize),
+            search: searchTerm,
+            ...(sorting[0] && {
+              sortBy: sorting[0].id,
+              sortDir: sorting[0].desc ? "desc" : "asc",
+            }),
+          })
+      ).then((r) => r.json())
+  );
+
+  return (
+    <div className="space-y-4">
+      <InputTypeIn
+        value={searchTerm}
+        onChange={(e) => setSearchTerm(e.target.value)}
+        placeholder="Search users..."
+      />
+      <DataTable
+        data={response?.items ?? []}
+        columns={columns}
+        getRowId={(row) => row.id}
+        searchTerm={searchTerm}
+        pageSize={pageSize}
+        footer={{ mode: "summary" }}
+        serverSide={{
+          totalItems: response?.total ?? 0,
+          isLoading,
+          onSortingChange: setSorting,
+          onPaginationChange: (idx, size) => {
+            setPageIndex(idx);
+            setPageSize(size);
+          },
+          onSearchTermChange: () => {
+            // search state is already managed above via searchTerm prop;
+            // this callback is useful for analytics or debouncing
+          },
+        }}
+      />
+    </div>
+  );
+}
+```
+
+## Sizing
+
+The `size` prop (`"regular"` or `"small"`) affects:
+
+- Qualifier column width (56px vs 40px)
+- Actions column width (88px vs 20px)
+- Footer text styles and pagination size
+- All child components via `TableSizeContext`
+
+Column widths can be responsive to size using a function:
+
+```ts
+// In types.ts, width accepts:
+width: ColumnWidth | ((size: TableSize) => ColumnWidth)
+
+// Example (this is what qualifier/actions use internally):
+width: (size) => size === "small" ? { fixed: 40 } : { fixed: 56 }
+```
+
+### Width system
+
+Data columns use **weight-based proportional distribution**. A column with `weight: 40` gets twice the space of one with `weight: 20`. When the container is narrower than the sum of `minWidth` values, columns clamp to their minimums.
+
+Fixed columns (`{ fixed: N }`) take exactly N pixels and don't participate in proportional distribution.
+
+Resizing uses **splitter semantics**: dragging a column border grows that column and shrinks its neighbor by the same amount, keeping total width constant.
+
+## Advanced Examples
+
+### Scrollable table with pinned header
+
+```tsx
+<DataTable
+  data={allRows}
+  columns={columns}
+  height={300}
+  headerBackground="var(--background-tint-00)"
+/>
+```
+
+### Hidden columns on load
+
+```tsx
+<DataTable
+  data={data}
+  columns={columns}
+  initialColumnVisibility={{ department: false, joinDate: false }}
+  footer={{ mode: "selection" }}
+/>
+```
+
+### Icon-based data column
+
+```tsx
+const STATUS_ICONS = {
+  active: SvgCheckCircle,
+  pending: SvgClock,
+  inactive: SvgAlertCircle,
+} as const;
+
+tc.column("status", {
+  header: "Status",
+  weight: 14,
+  minWidth: 80,
+  cell: (value) => (
+    <Content
+      sizePreset="main-ui"
+      variant="body"
+      icon={STATUS_ICONS[value]}
+      title={value.charAt(0).toUpperCase() + value.slice(1)}
+    />
+  ),
+})
+```
+
+### Non-selectable qualifier with icons
+
+```ts
+tc.qualifier({
+  content: "icon",
+  getIcon: (row) => row.icon,
+  selectable: false,
+  header: false,
+})
+```
+
+### Small variant in a bordered container
+
+```tsx
+<div className="border border-border-01 rounded-lg overflow-hidden">
+  <DataTable
+    data={data}
+    columns={columns}
+    size="small"
+    pageSize={10}
+    footer={{ mode: "selection" }}
+  />
+</div>
+```
+
+### Server-side pagination
+
+Minimal wiring for server-side mode — manage sorting/pagination state externally and pass the current page slice as `data`.
+
+```tsx
+<DataTable
+  data={currentPageRows}
+  columns={columns}
+  getRowId={(row) => row.id}
+  searchTerm={searchTerm}
+  pageSize={pageSize}
+  footer={{ mode: "summary" }}
+  serverSide={{
+    totalItems: totalCount,
+    isLoading,
+    onSortingChange: setSorting,
+    onPaginationChange: (idx, size) => {
+      setPageIndex(idx);
+      setPageSize(size);
+    },
+    onSearchTermChange: (term) => setSearchTerm(term),
+  }}
+/>
+```
+
+### Custom row click handler
+
+```tsx
+<DataTable
+  data={data}
+  columns={columns}
+  onRowClick={(row) => router.push(`/users/${row.id}`)}
+/>
+```
+
+## Source Files
+
+| File | Purpose |
+|---|---|
+| `DataTable.tsx` | Main component |
+| `columns.ts` | `createTableColumns` builder |
+| `types.ts` | All TypeScript interfaces |
+| `hooks/useDataTable.ts` | TanStack table wrapper hook |
+| `hooks/useColumnWidths.ts` | Weight-based width system |
+| `hooks/useDraggableRows.ts` | DnD hook (`@dnd-kit`) |
+| `Footer.tsx` | Selection / Summary footer modes |
+| `TableSizeContext.tsx` | Size context provider |

web/src/refresh-components/table/SortingPopover.tsx

@@ -21,7 +21,7 @@ import Text from "@/refresh-components/texts/Text";
 interface SortingPopoverProps<TData extends RowData = RowData> {
   table: Table<TData>;
   sorting: SortingState;
-  size?: "md" | "lg";
+  size?: "regular" | "small";
   footerText?: string;
   ascendingLabel?: string;
   descendingLabel?: string;
@@ -30,7 +30,7 @@ interface SortingPopoverProps<TData extends RowData = RowData> {
 function SortingPopover<TData extends RowData>({
   table,
   sorting,
-  size = "lg",
+  size = "regular",
   footerText,
   ascendingLabel = "Ascending",
   descendingLabel = "Descending",
@@ -48,8 +48,8 @@ function SortingPopover<TData extends RowData>({
         <Button
           icon={currentSort === null ? SvgArrowUpDown : SvgSortOrder}
           interaction={open ? "hover" : "rest"}
-          size={size === "md" ? "sm" : "md"}
-          prominence="tertiary"
+          size={size === "small" ? "sm" : "md"}
+          prominence="internal"
           tooltip="Sort"
         />
       </Popover.Trigger>
@@ -149,7 +149,7 @@ function SortingPopover<TData extends RowData>({
 // ---------------------------------------------------------------------------
 
 interface CreateSortingColumnOptions {
-  size?: "md" | "lg";
+  size?: "regular" | "small";
   footerText?: string;
   ascendingLabel?: string;
   descendingLabel?: string;

web/src/refresh-components/table/Table.stories.tsx

@@ -0,0 +1,281 @@
+import type { Meta, StoryObj } from "@storybook/react";
+import * as TooltipPrimitive from "@radix-ui/react-tooltip";
+import Table from "./Table";
+import TableHeader from "./TableHeader";
+import TableBody from "./TableBody";
+import TableRow from "./TableRow";
+import TableHead from "./TableHead";
+import TableCell from "./TableCell";
+import { TableSizeProvider } from "./TableSizeContext";
+import Text from "@/refresh-components/texts/Text";
+
+// ---------------------------------------------------------------------------
+// Meta
+// ---------------------------------------------------------------------------
+
+const meta: Meta<typeof Table> = {
+  title: "refresh-components/table/Table",
+  component: Table,
+  tags: ["autodocs"],
+  decorators: [
+    (Story) => (
+      <TooltipPrimitive.Provider>
+        <TableSizeProvider size="regular">
+          <div style={{ maxWidth: 800, padding: 16 }}>
+            <Story />
+          </div>
+        </TableSizeProvider>
+      </TooltipPrimitive.Provider>
+    ),
+  ],
+  parameters: {
+    layout: "padded",
+  },
+};
+
+export default meta;
+type Story = StoryObj<typeof Table>;
+
+// ---------------------------------------------------------------------------
+// Sample data
+// ---------------------------------------------------------------------------
+
+const connectors = [
+  {
+    name: "Google Drive",
+    type: "Cloud Storage",
+    docs: 1_240,
+    status: "Active",
+  },
+  { name: "Confluence", type: "Wiki", docs: 856, status: "Active" },
+  { name: "Slack", type: "Messaging", docs: 3_102, status: "Syncing" },
+  { name: "Notion", type: "Wiki", docs: 412, status: "Paused" },
+  { name: "GitHub", type: "Code", docs: 2_890, status: "Active" },
+];
+
+// ---------------------------------------------------------------------------
+// Stories
+// ---------------------------------------------------------------------------
+
+/** All primitive table components composed together (Table, TableHeader, TableBody, TableRow, TableHead, TableCell). */
+export const ComposedPrimitives: Story = {
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200}>Connector</TableHead>
+          <TableHead width={150}>Type</TableHead>
+          <TableHead width={120} alignment="right">
+            Documents
+          </TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c) => (
+          <TableRow key={c.name}>
+            <TableCell>
+              <Text mainUiBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMuted text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMono text03>
+                {c.docs.toLocaleString()}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};
+
+/** Table rows with the "table" variant (bottom border instead of rounded corners). */
+export const TableVariantRows: Story = {
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200}>Connector</TableHead>
+          <TableHead width={150}>Type</TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c) => (
+          <TableRow key={c.name} variant="table">
+            <TableCell>
+              <Text mainUiBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMuted text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};
+
+/** Row with selected state highlighted. */
+export const SelectedRows: Story = {
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200}>Connector</TableHead>
+          <TableHead width={150}>Type</TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c, i) => (
+          <TableRow key={c.name} selected={i === 1 || i === 3}>
+            <TableCell>
+              <Text mainUiBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMuted text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};
+
+/** Sortable table headers with sort indicators. */
+export const SortableHeaders: Story = {
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200} sorted="ascending" onSort={() => {}}>
+            Connector
+          </TableHead>
+          <TableHead width={150} sorted="none" onSort={() => {}}>
+            Type
+          </TableHead>
+          <TableHead width={120} sorted="descending" onSort={() => {}}>
+            Documents
+          </TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c) => (
+          <TableRow key={c.name}>
+            <TableCell>
+              <Text mainUiBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMuted text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMono text03>
+                {c.docs.toLocaleString()}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};
+
+/** Small size variant with denser spacing. */
+export const SmallSize: Story = {
+  decorators: [
+    (Story) => (
+      <TooltipPrimitive.Provider>
+        <TableSizeProvider size="small">
+          <div style={{ maxWidth: 800, padding: 16 }}>
+            <Story />
+          </div>
+        </TableSizeProvider>
+      </TooltipPrimitive.Provider>
+    ),
+  ],
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200}>Connector</TableHead>
+          <TableHead width={150}>Type</TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c) => (
+          <TableRow key={c.name}>
+            <TableCell>
+              <Text secondaryBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text secondaryBody text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text secondaryBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};
+
+/** Disabled rows styling. */
+export const DisabledRows: Story = {
+  render: () => (
+    <Table>
+      <TableHeader>
+        <TableRow>
+          <TableHead width={200}>Connector</TableHead>
+          <TableHead width={150}>Type</TableHead>
+          <TableHead width={120}>Status</TableHead>
+        </TableRow>
+      </TableHeader>
+      <TableBody>
+        {connectors.map((c, i) => (
+          <TableRow key={c.name} disabled={i === 2 || i === 4}>
+            <TableCell>
+              <Text mainUiBody>{c.name}</Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiMuted text03>
+                {c.type}
+              </Text>
+            </TableCell>
+            <TableCell>
+              <Text mainUiBody>{c.status}</Text>
+            </TableCell>
+          </TableRow>
+        ))}
+      </TableBody>
+    </Table>
+  ),
+};

web/src/refresh-components/table/Table.tsx

@@ -0,0 +1,26 @@
+import { cn } from "@/lib/utils";
+import type { WithoutStyles } from "@/types";
+
+interface TableProps
+  extends WithoutStyles<React.TableHTMLAttributes<HTMLTableElement>> {
+  ref?: React.Ref<HTMLTableElement>;
+  /** Explicit pixel width for the table (e.g. from `table.getTotalSize()`).
+   *  When provided the table uses exactly this width instead of stretching
+   *  to fill its container, which prevents `table-layout: fixed` from
+   *  redistributing extra space across columns on resize. */
+  width?: number;
+}
+
+function Table({ ref, width, ...props }: TableProps) {
+  return (
+    <table
+      ref={ref}
+      className={cn("border-separate border-spacing-0", "min-w-full")}
+      style={{ tableLayout: "fixed", width: width ?? undefined }}
+      {...props}
+    />
+  );
+}
+
+export default Table;
+export type { TableProps };

web/src/refresh-components/table/TableCell.tsx

@@ -1,6 +1,6 @@
-import { cn } from "@opal/utils";
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { cn } from "@/lib/utils";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import type { WithoutStyles } from "@/types";
 
 interface TableCellProps

web/src/refresh-components/table/TableHead.tsx

@@ -1,7 +1,7 @@
-import { cn } from "@opal/utils";
+import { cn } from "@/lib/utils";
 import Text from "@/refresh-components/texts/Text";
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import type { WithoutStyles } from "@/types";
 import { Button } from "@opal/components";
 import { SvgChevronDown, SvgChevronUp, SvgHandle, SvgSort } from "@opal/icons";
@@ -30,7 +30,7 @@ interface TableHeadCustomProps {
   icon?: (sorted: SortDirection) => IconFunctionComponent;
   /** Text alignment for the column. Defaults to `"left"`. */
   alignment?: "left" | "center" | "right";
-  /** Cell density. `"md"` uses tighter padding for denser layouts. */
+  /** Cell density. `"small"` uses tighter padding for denser layouts. */
   size?: TableSize;
   /** Column width in pixels. Applied as an inline style on the `<th>`. */
   width?: number;
@@ -88,7 +88,7 @@ export default function TableHead({
 }: TableHeadProps) {
   const contextSize = useTableSize();
   const resolvedSize = size ?? contextSize;
-  const isSmall = resolvedSize === "md";
+  const isSmall = resolvedSize === "small";
   return (
     <th
       {...thProps}

web/src/refresh-components/table/TableQualifier.tsx

@@ -1,12 +1,12 @@
 "use client";
 
 import React from "react";
-import { cn } from "@opal/utils";
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { cn } from "@/lib/utils";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import { SvgUser } from "@opal/icons";
 import type { IconFunctionComponent } from "@opal/types";
-import type { QualifierContentType } from "./types";
+import type { QualifierContentType } from "@/refresh-components/table/types";
 import Checkbox from "@/refresh-components/inputs/Checkbox";
 import Text from "@/refresh-components/texts/Text";
 
@@ -35,8 +35,8 @@ interface TableQualifierProps {
 }
 
 const iconSizes = {
-  lg: 16,
-  md: 14,
+  regular: 16,
+  small: 14,
 } as const;
 
 function getQualifierStyles(selected: boolean, disabled: boolean) {
@@ -115,7 +115,7 @@ function TableQualifier({
           <div
             className={cn(
               "flex items-center justify-center rounded-full bg-background-neutral-inverted-00",
-              resolvedSize === "lg" ? "h-7 w-7" : "h-6 w-6"
+              resolvedSize === "regular" ? "h-7 w-7" : "h-6 w-6"
             )}
           >
             <Text
@@ -138,7 +138,7 @@ function TableQualifier({
     <div
       className={cn(
         "group relative inline-flex shrink-0 items-center justify-center",
-        resolvedSize === "lg" ? "h-9 w-9" : "h-7 w-7",
+        resolvedSize === "regular" ? "h-9 w-9" : "h-7 w-7",
         disabled ? "cursor-not-allowed" : "cursor-default",
         className
       )}
@@ -147,7 +147,7 @@ function TableQualifier({
       <div
         className={cn(
           "flex items-center justify-center overflow-hidden transition-colors",
-          resolvedSize === "lg" ? "h-9 w-9" : "h-7 w-7",
+          resolvedSize === "regular" ? "h-9 w-9" : "h-7 w-7",
           isRound ? "rounded-full" : "rounded-08",
           styles.container,
           content === "image" && disabled && !selected && "opacity-50"

web/src/refresh-components/table/TableRow.tsx

@@ -1,8 +1,8 @@
 "use client";
 
-import { cn } from "@opal/utils";
-import { useTableSize } from "./TableSizeContext";
-import type { TableSize } from "./TableSizeContext";
+import { cn } from "@/lib/utils";
+import { useTableSize } from "@/refresh-components/table/TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import type { WithoutStyles } from "@/types";
 import { useSortable } from "@dnd-kit/sortable";
 import { CSS } from "@dnd-kit/utilities";
@@ -95,7 +95,7 @@ function SortableTableRow({
             {...listeners}
           >
             <SvgHandle
-              size={resolvedSize === "md" ? 12 : 16}
+              size={resolvedSize === "small" ? 12 : 16}
               className="text-border-02"
             />
           </button>

web/src/refresh-components/table/TableSizeContext.tsx

@@ -1,11 +1,10 @@
 "use client";
 
 import { createContext, useContext } from "react";
-import type { SizeVariants } from "@opal/types";
 
-type TableSize = Extract<SizeVariants, "md" | "lg">;
+type TableSize = "regular" | "small";
 
-const TableSizeContext = createContext<TableSize>("lg");
+const TableSizeContext = createContext<TableSize>("regular");
 
 interface TableSizeProviderProps {
   size: TableSize;

web/src/refresh-components/table/columns.ts

@@ -13,10 +13,10 @@ import type {
   OnyxDataColumn,
   OnyxDisplayColumn,
   OnyxActionsColumn,
-} from "./types";
-import type { TableSize } from "./TableSizeContext";
+} from "@/refresh-components/table/types";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import type { IconFunctionComponent } from "@opal/types";
-import type { SortDirection } from "./TableHead";
+import type { SortDirection } from "@/refresh-components/table/TableHead";
 
 // ---------------------------------------------------------------------------
 // Qualifier column config
@@ -160,7 +160,7 @@ export function createTableColumns<TData>(): TableColumnsBuilder<TData> {
         id: "qualifier",
         def,
         width: (size: TableSize) =>
-          size === "md" ? { fixed: 40 } : { fixed: 56 },
+          size === "small" ? { fixed: 40 } : { fixed: 56 },
         content,
         headerContentType: config?.headerContentType,
         getInitials: config?.getInitials,
@@ -246,7 +246,7 @@ export function createTableColumns<TData>(): TableColumnsBuilder<TData> {
         id: "__actions",
         def,
         width: (size: TableSize) =>
-          size === "md" ? { fixed: 20 } : { fixed: 88 },
+          size === "small" ? { fixed: 20 } : { fixed: 88 },
         showColumnVisibility: config?.showColumnVisibility ?? true,
         showSorting: config?.showSorting ?? true,
         sortingFooterText: config?.sortingFooterText,

web/src/refresh-components/table/types.ts

@@ -4,9 +4,9 @@ import type {
   SortingState,
   VisibilityState,
 } from "@tanstack/react-table";
-import type { TableSize } from "./TableSizeContext";
+import type { TableSize } from "@/refresh-components/table/TableSizeContext";
 import type { IconFunctionComponent } from "@opal/types";
-import type { SortDirection } from "./TableHead";
+import type { SortDirection } from "@/refresh-components/table/TableHead";
 
 // ---------------------------------------------------------------------------
 // Column width (mirrors useColumnWidths types)
@@ -166,7 +166,7 @@ export interface DataTableProps<TData> {
   draggable?: DataTableDraggableConfig;
   /** Footer configuration. */
   footer?: DataTableFooterConfig;
-  /** Table size variant. @default "lg" */
+  /** Table size variant. @default "regular" */
   size?: TableSize;
   /** Called whenever the set of selected row IDs changes. Receives IDs produced by `getRowId`. */
   onSelectionChange?: (selectedIds: string[]) => void;

web/src/refresh-pages/AppPage.tsx

@@ -21,6 +21,7 @@ import useTags from "@/hooks/useTags";
 import { useDocumentSets } from "@/lib/hooks/useDocumentSets";
 import { useAgents } from "@/hooks/useAgents";
 import { AppPopup } from "@/app/app/components/AppPopup";
+import ExceptionTraceModal from "@/components/modals/ExceptionTraceModal";
 import { useUser } from "@/providers/UserProvider";
 import NoAgentModal from "@/components/modals/NoAgentModal";
 import PreviewModal from "@/sections/modals/PreviewModal";
@@ -419,6 +420,10 @@ export default function AppPage({ firstMessage }: ChatPageProps) {
     }
   }, [currentChatSessionId]);
 
+  const [stackTraceModalContent, setStackTraceModalContent] = useState<
+    string | null
+  >(null);
+
   const handleResubmitLastMessage = useCallback(() => {
     // Grab the last user-type message
     const lastUserMsg = messageHistory
@@ -655,6 +660,13 @@ export default function AppPage({ firstMessage }: ChatPageProps) {
         />
       )}
 
+      {stackTraceModalContent && (
+        <ExceptionTraceModal
+          onOutsideClick={() => setStackTraceModalContent(null)}
+          exceptionTrace={stackTraceModalContent}
+        />
+      )}
+
       <FederatedOAuthModal />
 
       <AppLayouts.Root enableBackground={!appFocus.isProject()}>

web/src/refresh-pages/admin/LLMConfigurationPage.tsx

@@ -35,16 +35,16 @@ import {
   WellKnownLLMProviderDescriptor,
 } from "@/interfaces/llm";
 import { getModalForExistingProvider } from "@/sections/modals/llmConfig/getModal";
-import OpenAIModal from "@/sections/modals/llmConfig/OpenAIModal";
-import AnthropicModal from "@/sections/modals/llmConfig/AnthropicModal";
-import OllamaModal from "@/sections/modals/llmConfig/OllamaModal";
-import AzureModal from "@/sections/modals/llmConfig/AzureModal";
-import BedrockModal from "@/sections/modals/llmConfig/BedrockModal";
-import VertexAIModal from "@/sections/modals/llmConfig/VertexAIModal";
-import OpenRouterModal from "@/sections/modals/llmConfig/OpenRouterModal";
-import CustomModal from "@/sections/modals/llmConfig/CustomModal";
-import LMStudioForm from "@/sections/modals/llmConfig/LMStudioForm";
-import LiteLLMProxyModal from "@/sections/modals/llmConfig/LiteLLMProxyModal";
+import { OpenAIModal } from "@/sections/modals/llmConfig/OpenAIModal";
+import { AnthropicModal } from "@/sections/modals/llmConfig/AnthropicModal";
+import { OllamaModal } from "@/sections/modals/llmConfig/OllamaModal";
+import { AzureModal } from "@/sections/modals/llmConfig/AzureModal";
+import { BedrockModal } from "@/sections/modals/llmConfig/BedrockModal";
+import { VertexAIModal } from "@/sections/modals/llmConfig/VertexAIModal";
+import { OpenRouterModal } from "@/sections/modals/llmConfig/OpenRouterModal";
+import { CustomModal } from "@/sections/modals/llmConfig/CustomModal";
+import { LMStudioForm } from "@/sections/modals/llmConfig/LMStudioForm";
+import { LiteLLMProxyModal } from "@/sections/modals/llmConfig/LiteLLMProxyModal";
 import { Section } from "@/layouts/general-layouts";
 
 const route = ADMIN_ROUTES.LLM_MODELS;
@@ -53,20 +53,6 @@ const route = ADMIN_ROUTES.LLM_MODELS;
 // Provider form mapping (keyed by provider name from the API)
 // ============================================================================
 
-// Client-side ordering for the "Add Provider" cards. The backend may return
-// wellKnownLLMProviders in an arbitrary order, so we sort explicitly here.
-const PROVIDER_DISPLAY_ORDER: string[] = [
-  "openai",
-  "anthropic",
-  "vertex_ai",
-  "bedrock",
-  "azure",
-  "litellm_proxy",
-  "ollama_chat",
-  "openrouter",
-  "lm_studio",
-];
-
 const PROVIDER_MODAL_MAP: Record<
   string,
   (
@@ -470,32 +456,23 @@ export default function LLMConfigurationPage() {
           />
 
           <div className="grid grid-cols-2 gap-2">
-            {[...(wellKnownLLMProviders ?? [])]
-              .sort((a, b) => {
-                const aIndex = PROVIDER_DISPLAY_ORDER.indexOf(a.name);
-                const bIndex = PROVIDER_DISPLAY_ORDER.indexOf(b.name);
-                return (
-                  (aIndex === -1 ? Infinity : aIndex) -
-                  (bIndex === -1 ? Infinity : bIndex)
+            {wellKnownLLMProviders?.map((provider) => {
+              const formFn = PROVIDER_MODAL_MAP[provider.name];
+              if (!formFn) {
+                toast.error(
+                  `No modal mapping for provider "${provider.name}".`
                 );
-              })
-              .map((provider) => {
-                const formFn = PROVIDER_MODAL_MAP[provider.name];
-                if (!formFn) {
-                  toast.error(
-                    `No modal mapping for provider "${provider.name}".`
-                  );
-                  return null;
-                }
-                return (
-                  <NewProviderCard
-                    key={provider.name}
-                    provider={provider}
-                    isFirstProvider={isFirstProvider}
-                    formFn={formFn}
-                  />
-                );
-              })}
+                return null;
+              }
+              return (
+                <NewProviderCard
+                  key={provider.name}
+                  provider={provider}
+                  isFirstProvider={isFirstProvider}
+                  formFn={formFn}
+                />
+              );
+            })}
             <NewCustomProviderCard isFirstProvider={isFirstProvider} />
           </div>
         </GeneralLayouts.Section>

web/src/refresh-pages/admin/UsersPage/UserRowActions.tsx

@@ -12,7 +12,6 @@ import {
   SvgKey,
 } from "@opal/icons";
 import { Disabled } from "@opal/core";
-import LineItem from "@/refresh-components/buttons/LineItem";
 import Popover from "@/refresh-components/Popover";
 import Separator from "@/refresh-components/Separator";
 import { Section } from "@/layouts/general-layouts";
@@ -79,17 +78,18 @@ export default function UserRowActions({
       return (
         <>
           {user.id && (
-            <LineItem
+            <Button
+              prominence="tertiary"
               icon={SvgUsers}
               onClick={() => openModal(Modal.EDIT_GROUPS)}
             >
               Groups &amp; Roles
-            </LineItem>
+            </Button>
           )}
           <Disabled disabled>
-            <LineItem danger icon={SvgUserX}>
+            <Button prominence="tertiary" variant="danger" icon={SvgUserX}>
               Deactivate User
-            </LineItem>
+            </Button>
           </Disabled>
           <Separator paddingXRem={0.5} />
           <Text as="p" secondaryBody text03 className="px-3 py-1">
@@ -102,18 +102,20 @@ export default function UserRowActions({
     switch (user.status) {
       case UserStatus.INVITED:
         return (
-          <LineItem
-            danger
+          <Button
+            prominence="tertiary"
+            variant="danger"
             icon={SvgXCircle}
             onClick={() => openModal(Modal.CANCEL_INVITE)}
           >
             Cancel Invite
-          </LineItem>
+          </Button>
         );
 
       case UserStatus.REQUESTED:
         return (
-          <LineItem
+          <Button
+            prominence="tertiary"
             icon={SvgUserCheck}
             onClick={() => {
               setPopoverOpen(false);
@@ -131,34 +133,37 @@ export default function UserRowActions({
             }}
           >
             Approve
-          </LineItem>
+          </Button>
         );
 
       case UserStatus.ACTIVE:
         return (
           <>
             {user.id && (
-              <LineItem
+              <Button
+                prominence="tertiary"
                 icon={SvgUsers}
                 onClick={() => openModal(Modal.EDIT_GROUPS)}
               >
                 Groups &amp; Roles
-              </LineItem>
+              </Button>
             )}
-            <LineItem
+            <Button
+              prominence="tertiary"
               icon={SvgKey}
               onClick={() => openModal(Modal.RESET_PASSWORD)}
             >
               Reset Password
-            </LineItem>
+            </Button>
             <Separator paddingXRem={0.5} />
-            <LineItem
-              danger
+            <Button
+              prominence="tertiary"
+              variant="danger"
               icon={SvgUserX}
               onClick={() => openModal(Modal.DEACTIVATE)}
             >
               Deactivate User
-            </LineItem>
+            </Button>
           </>
         );
 
@@ -166,34 +171,38 @@ export default function UserRowActions({
         return (
           <>
             {user.id && (
-              <LineItem
+              <Button
+                prominence="tertiary"
                 icon={SvgUsers}
                 onClick={() => openModal(Modal.EDIT_GROUPS)}
               >
                 Groups &amp; Roles
-              </LineItem>
+              </Button>
             )}
-            <LineItem
+            <Button
+              prominence="tertiary"
               icon={SvgKey}
               onClick={() => openModal(Modal.RESET_PASSWORD)}
             >
               Reset Password
-            </LineItem>
+            </Button>
             <Separator paddingXRem={0.5} />
-            <LineItem
+            <Button
+              prominence="tertiary"
               icon={SvgUserPlus}
               onClick={() => openModal(Modal.ACTIVATE)}
             >
               Activate User
-            </LineItem>
+            </Button>
             <Separator paddingXRem={0.5} />
-            <LineItem
-              danger
+            <Button
+              prominence="tertiary"
+              variant="danger"
               icon={SvgUserX}
               onClick={() => openModal(Modal.DELETE)}
             >
               Delete User
-            </LineItem>
+            </Button>
           </>
         );
 

web/src/refresh-pages/admin/UsersPage/UsersTable.tsx

@@ -1,7 +1,8 @@
 "use client";
 
 import { useMemo, useState } from "react";
-import { Table, createTableColumns } from "@opal/components";
+import DataTable from "@/refresh-components/table/DataTable";
+import { createTableColumns } from "@/refresh-components/table/columns";
 import { Content } from "@opal/layouts";
 import { Button } from "@opal/components";
 import { SvgDownload } from "@opal/icons";
@@ -215,7 +216,7 @@ export default function UsersTable({
         roleCounts={roleCounts}
         statusCounts={statusCounts}
       />
-      <Table
+      <DataTable
         data={filteredUsers}
         columns={columns}
         getRowId={(row) => row.id ?? row.email}