Summary

Severity: Medium-High
Area: Persistence / API
Affected files: agentkit/api/schemas.py:9-18, agentkit/db/models.py:25-29
Suggested branch: fix/session-id-length-validation
Suggested PR title: fix(api): bound session_id at the request boundary

This is finding AUDIT-09 of 18 from a consolidated audit of agentkit/. It is deliberately self-contained so it can be fixed on its own branch with its own PR.

Related findings: Related: AUDIT-08.

AUDIT-09 — Medium-High — session_id unbounded at the API, VARCHAR(255) in the DB

Files

• agentkit/api/schemas.py:9-18
• agentkit/db/models.py:25-29

Current code — agentkit/api/schemas.py:9-18

class RunCreateRequest(BaseModel):
    """Payload to initiate a new agent execution run."""

    model_config = ConfigDict(extra="forbid")

    goal: str = Field(..., min_length=1, description="Goal or task description for the agent.")
    session_id: str | None = Field(
        default=None,
        description="Optional conversation session ID to preserve context across multiple runs.",
    )

Current code — agentkit/db/models.py:25-29

    session_id: Mapped[str | None] = mapped_column(
        String(255),
        nullable=True,
        index=True,
    )

Why this is a bug

The API accepts a session_id of any length; the column is VARCHAR(255). A 256+ character session id makes the INSERT in create_run fail with a data-length error → suppressed (AUDIT-08) → the client receives a phantom HTTP 200 succeeded and the run is never stored. The same id is also used verbatim as a Redis key (session:<id>:messages), so an oversized id also becomes an unbounded key.

Pass cases

• RunCreateRequest.session_id has an explicit maximum length (255) → an over-long id returns HTTP 422 with the validation error, before any agent work starts.
• An over-long session id never reaches the database or Redis.
• A 255-character session id is accepted and round-trips (DB + Redis).
• The DB column length and the API limit are asserted to be equal by a test (so they cannot drift again).
• A test asserts an over-long session id yields 422 and no run record is created.

Definition of done (per AGENTS.md)

• New/updated unit and integration tests covering the pass cases above, all passing (pytest).
• ruff check ., ruff format --check . and mypy agentkit clean.
• Docstrings updated where behaviour changed; docs/PROGRESS.md updated (status, PR link, commit count, notes).
• A new ADR in docs/DECISIONS.md if an architectural decision was made.
• Branch pushed and PR opened (never commit directly to main).

Baseline at the time of writing: pytest is green on main (all tests pass; tests/real/* skipped unless AGENTKIT_REAL_INFRA=1).