Compare an installed package, its release archive, and its source repository; distinguish source drift from runtime state; and backport verified installed-only fixes. Use when installed files or a ZIP were edited directly, or when the repository may be stale.
Maintain product, architecture, design, API, data, and operational documentation in an established repository. Use when deciding whether to edit an existing source of truth or add a document, separating current-state contracts from implementation plans,…
Design, implement, and review CI test-result parsers, PR comments, job summaries, artifacts, and required checks. Use when cancelled or timed-out runs are misreported as test failures, setup/global errors disappear because zero tests ran, diagnostics may…
Locate, identify, and safely summarize prior local Codex conversations from session JSONL when a user asks whether an earlier task, decision, or implementation can be recovered.
Investigate GitHub Actions runtime regressions by comparing recent runs, job and step durations, test counts, slow files, exact-SHA repeats, and critical-path effects. Use when CI or E2E feels slower, a suspected cause needs falsification, or optimization…
Diagnose and fix Playwright tests where a lazy per-test fixture is created after a beforeEach navigation, leaving the page with stale source data. Use when DB fixture prechecks pass but the browser cannot find the created entity.
Use for Codex models/pricing, scheduled tasks, skills, settings, setup, troubleshooting, customization, automations, and self-knowledge—including 'you,' 'your,' 'this app,' or 'this coding agent' when they refer to Codex—and for OpenAI APIs/products and…
Integrate a new or migrated UI into an established product so its layout, surfaces, states, and interactions feel native without coupling unrelated domain logic. Use when a page looks stylistically different from neighboring workflows, when deciding what…