codex/job-timeout-contract

[alain-sv]

No description provided.

[qodo-code-review]

Review Summary by Qodo

Expose job timeout and release notes metadata for Studio integration

✨ Enhancement 🧪 Tests

Walkthroughs

Description

• Expose job timeout metadata in AgentMethod and registration contract
  - Add timeout field (defaults to 600 seconds, nullable for manual control)
  - Include is_async in registration info for Studio consumption
• Add release_notes_url field to Agent and registration models
  - Optional URL field for agent version release notes
  - Propagated through AgentAbstract, AgentResponse, and AgentRegistrationContract
• Update dependencies and development tooling
  - Bump runtime and dev dependencies to latest versions
  - Add Black formatter and yamllint to CI pipeline
• Add comprehensive test coverage for new metadata fields

Diagram

flowchart LR
  AM["AgentMethod"] -->|adds timeout field| RI["registration_info"]
  AA["AgentAbstract"] -->|adds release_notes_url| AR["AgentResponse"]
  ARC["AgentRegistrationContract"] -->|includes| RNU["release_notes_url"]
  AMC["AgentMethodContract"] -->|includes| TM["timeout + is_async"]
  RI -->|exposes to| Studio["Studio"]
  AR -->|exposes to| Studio

Loading

File Changes

1. src/supervaizer/agent.py ✨ Enhancement +18/-0

Add timeout and release notes metadata to agent models

src/supervaizer/agent.py

---

2. src/supervaizer/contracts.py ✨ Enhancement +3/-0

Update contracts with timeout and release notes fields

src/supervaizer/contracts.py

---

3. tests/test_agent.py 🧪 Tests +37/-0

Add tests for timeout and release notes metadata

tests/test_agent.py

---

View more (7)

4. tests/test_contracts.py 🧪 Tests +10/-0

Verify timeout and release notes in contract schema

tests/test_contracts.py

---

5. tests/test_cli.py 🧪 Tests +9/-0

Add console fixture for consistent test output

tests/test_cli.py

---

6. pyproject.toml Dependencies +40/-20

Bump dependencies and add Black formatter config

pyproject.toml

---

7. .github/workflows/python-package.yml ⚙️ Configuration changes +7/-0

Add Black and yamllint checks to CI pipeline

.github/workflows/python-package.yml

---

8. .claude/skills/gitnexus/gitnexus-cli/SKILL.md 📝 Documentation +1/-0

Document drop-embeddings flag for gitnexus analyze

.claude/skills/gitnexus/gitnexus-cli/SKILL.md

---

9. AGENTS.md 📝 Documentation +17/-73

Update GitNexus index statistics and documentation

AGENTS.md

---

10. docs/CHANGELOG.md 📝 Documentation +21/-0

Document timeout metadata and dependency updates

docs/CHANGELOG.md

---

[qodo-code-review]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)

1. Custom async flag misread ✓ Resolved 🐞 Bug ≡ Correctness

Description

Agent.job_start() selects a custom action when method_name != "job_start", but it still checks
self.methods.job_start.is_async to decide sync/async execution; this can execute an async custom
method via the synchronous _execute() path and raise TypeError when the coroutine is not a
JobResponse. This becomes user-visible because the PR now advertises per-method is_async in the
Studio contract/registration payload.

Code

src/supervaizer/contracts.py[R156-157]

+    is_async: bool = False
+    timeout: int | None = 600

Evidence

The PR now exposes per-method is_async in the contract/registration payload, but the runtime
dispatch in Agent.job_start() checks the job_start method’s async flag even after selecting a
custom method as action. Since _execute() calls the method directly and requires a
JobResponse, invoking an async method through this sync path produces a coroutine and triggers the
TypeError guard.

src/supervaizer/contracts.py[150-158]
src/supervaizer/agent.py[453-467]
src/supervaizer/agent.py[916-950]
src/supervaizer/agent.py[865-882]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The controller runtime decides whether to execute a method synchronously vs asynchronously using `self.methods.job_start.is_async` even when the selected method is a custom action. Now that the registration/contract exposes per-method `is_async`, Studio (or agent authors) can mark a custom method as async, but the controller may still execute it synchronously, causing `_execute()` to receive a coroutine and raise because it is not a `JobResponse`.
### Issue Context
- PR adds `is_async` to the contract and exports it in method registration info.
- `Agent.job_start()` already chooses `action = self.methods.custom[method_name]` for custom methods, but it does not use `action.is_async`.
### Fix Focus Areas
- src/supervaizer/agent.py[916-950]
- src/supervaizer/agent.py[865-882]
- src/supervaizer/contracts.py[150-158]
### Proposed fix
1. In `Agent.job_start()`, change the conditional from `if self.methods.job_start.is_async:` to `if action.is_async:`.
2. (Optional but recommended) Adjust the initial log line that currently prints `self.methods.job_start.method` so it logs `action.method` (so custom invocations log correctly).
3. Add/extend a unit test that constructs an agent with a custom method where `is_async=True` and asserts the controller takes the async branch (currently `NotImplementedError` is fine, but it should be raised based on `action.is_async`, not `job_start.is_async`).

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Formatter enforcement mismatch ✓ Resolved 🐞 Bug ☼ Reliability

Description

CI now runs black --check, but local pre-commit formatting is driven by ruff format and does
not run Black, so a change can pass local hooks yet fail CI due to unmet Black formatting. This
creates an inconsistent single source of truth for formatting.

Code

.github/workflows/python-package.yml[R31-32]

+      - name: Check Black formatting
+        run: uv run black --check .

Evidence

The workflow explicitly adds Black enforcement, while the repository’s pre-commit hooks format using
Ruff and do not include Black. That means local formatting compliance is not equivalent to CI
formatting compliance.

.github/workflows/python-package.yml[24-37]
.pre-commit-config.yaml[1-14]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The workflow introduces a mandatory Black check in CI, but the repository’s pre-commit configuration formats Python with Ruff (`ruff format`) and does not run Black. This makes it possible to pass pre-commit locally and still fail CI on formatting.
### Issue Context
- CI: `uv run black --check .`
- Local hooks: `uv run ruff format --force-exclude` (no Black hook)
### Fix Focus Areas
- .github/workflows/python-package.yml[24-37]
- .pre-commit-config.yaml[1-15]
- pyproject.toml[116-135]
### Suggested resolutions (pick one)
1. **Make Black the single formatter**: add a Black hook to `.pre-commit-config.yaml` (using `uv run black ...`) so local and CI enforce the same formatter.
2. **Make Ruff the single formatter**: remove the Black CI check step and rely on Ruff formatting checks (or add a Ruff check step instead).
3. If keeping both, add documentation + a `just format`/`just lint` target that runs both so developers can reproduce CI locally.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---