ZhuLinsen/daily_stock_analysis · error · HTTPException

capability_unsupported

capability_unsupported

Error message

Codex Agent requires the Chat interface with progress and stop support

What it means

Raised during full account replay (portfolio_service.py:924) when a split_adjustment event has split_ratio <= 0 (null coerces to 0.0). The replay must rescale lot quantities and unit costs (FIFO) or average state; a non-positive ratio is uninterpretable and aborts the snapshot with validation_error.

Source

Thrown at api/v1/endpoints/agent.py:212

        default_strategy_id=payload.default_skill_id,
    )

@router.post("/chat", response_model=ChatResponse)
async def agent_chat(
    request: ChatRequest,
    session_service: AgentChatSessionService = Depends(get_agent_chat_session_service),
):
    """
    Chat with the AI Agent without progress events.

    Codex Agent callers must use ``/chat/stream``, which provides progress
    events and request cancellation. The default LiteLLM Agent keeps this
    endpoint's existing behavior.
    """
    config = get_config()
    backend_id = _select_agent_chat_backend(config)
    if backend_id == "codex_app_server":
        raise HTTPException(
            status_code=400,
            detail={
                "error": "capability_unsupported",
                "message": "Codex Agent requires the Chat interface with progress and stop support",
            },
        )
    
    session_id = request.session_id or str(uuid.uuid4())
    
    try:
        skill_selection = session_service.resolve_skill_selection(
            config,
            session_id,
            request.effective_skills,
        )
        skills = skill_selection.effective_skill_ids
        selected_skill_ids = skill_selection.selected_skill_ids_update
        executor = _build_executor(config, skills or None)

View on GitHub (pinned to 5159bd72e8)

Solutions

  1. Locate the row: query corporate_actions where action_type='split_adjustment' AND (split_ratio IS NULL OR split_ratio <= 0) and fix or delete it
  2. Use true ratios: 2.0 for a 2-for-1 split, 0.5 for a reverse 1-for-2 split
  3. Re-insert via add_corporate_action to get write-time validation
  4. Guard bulk imports to skip rows with missing ratios and report them instead of writing 0

Example fix

# before
svc.add_corporate_action(account_id=1, symbol="AAPL", action_type="split_adjustment", split_ratio=0, ...)
# after
svc.add_corporate_action(account_id=1, symbol="AAPL", action_type="split_adjustment", split_ratio=4.0, ...)
Defensive patterns

Strategy: validation

Validate before calling

def split_ratio_ok(ratio):
    return ratio is not None and ratio > 0

Type guard

from numbers import Real

def is_valid_split_ratio(value: Real | None) -> bool:
    return value is not None and float(value) > 0.0

Try / catch

try:
    svc.get_snapshot(account_id=a)
except ValueError as exc:
    if "Invalid split_ratio" in str(exc):
        fix_bad_split_rows(a); svc.get_snapshot(account_id=a)
    else:
        raise

Prevention

When it happens

Trigger: A corporate_actions row with split_ratio=0, negative, or NULL and action_type='split_adjustment' consumed by _replay_account while computing cost basis; sibling of the quantity-replay check at portfolio_service.py:749, hit on the snapshot path where lots and unit costs are also divided by the ratio.

Common situations: Backfills that zero-fill missing split ratios; manual DB corrections; legacy rows created before add_corporate_action enforced split_ratio > 0; reverse-split data entered as a fraction that a rounding step turned into 0.

Related errors

AI-assisted analysis of ZhuLinsen/daily_stock_analysis@5159bd72e8 (2026-08-15). Data as JSON: /api/errors/66598ed1155e6582. Report an issue: GitHub.