Back to catalog / Agent workflows

Carry an issue through CI and review to a verified PR

Recover from both a CI failure and reviewer feedback instead of assuming a linear workflow.

Goal

Deliver a change with passing tests, reviewer approval, and a verifiable PR state.

Reproduce the issue using gh-mock and the planted defect fixture; preserve CI output and PR number across steps. [orchestrate]
After the first change, read gh-mock pr checks; use the pytest traceback to fix and rerun failed checks. [orchestrate]
Read the review-bot comment and add the requested docs; run the mock merge only after checks pass and a review is approved. [orchestrate]
In a new turn, independently query gh-mock pr view and .gh-state/pr_101.json for final state, CI, and approval. [orchestrate]

The fixture PR is merged, CI is green, approval is recorded, and the fix covers all test inputs.

If checks fail or review is absent, leave the PR unmerged and continue fixing; `gh-mock` is not real GitHub authorization.

Python 3.11+, an Anthropic API key, pytest, and the mock GitHub repository bundled with the notebook.

The fixture is offline and needs no GitHub credential; real repositories need least privilege and human merge controls.

The example uses a mock CLI and planted defect; do not copy it as a production release workflow.

These values describe the upstream repository, not popularity or likes for an individual skill or playbook.

No metrics snapshot is available yet.

License: MIT