AI Agent Workflow Collection

Ensure ADRs capture context, decision, alternatives, tradeoffs, and follow-ups in a way that prevents future drift.

Review changes to the Agent Skills import/export bridge for path safety, round-trip fidelity, and validation correctness.

Checks whether AI-assisted work defines review criteria, accessible context, explicit review loop, human judgment boundary, and feedback capture before delegating to an agent.

Detects per-caller special-cases (bandaids) bolted onto shared infrastructure/common functions, and when two or more same-kind special-cases exist proposes generalizing the lower-level mechanism instead.

Detect breaking API contract changes, DTO modifications without compatibility handling, and missing tests for changed API boundaries.

Ensure API/contract changes specify versioning strategy, backward compatibility, deprecation plan, and migration guidance.

Ensure architecture updates document affected stakeholders, notification plan, and deprecation/retirement signals to keep knowledge aligned.

Ensure design docs explicitly capture risks, assumptions, and open questions with owners, deadlines, and mitigation plans.

Ensure design changes stay consistent across ADRs, diagrams, and specs; decisions are traceable; and drift is explicitly managed.

Detect missing validation plans (how to verify the design is correct) in design documents and ADRs.

Ensure architecture docs capture availability targets, failover strategy, capacity headroom, and resilience trade-offs for critical services.

Ensure architecture docs define bounded contexts, ownership, and a consistent ubiquitous language to prevent domain drift and coupling.

Detect undefined or inconsistent cache strategies (layers, consistency, invalidation, TTL, failure handling) in design documents.

Ensure architecture/design docs state traffic assumptions, performance budgets, resource limits, and cost risks for critical paths.

Detects long-lived objects (caches, listeners, singletons) that capture an entire enclosing scope via closures/environment capture, keeping large arrays or buffers alive, and proposes copying only the needed fields.

Find missing tests for critical paths, edge cases, and failure handling in changed code.

Ensure designs define data flow, state ownership, consistency boundaries, and cross-boundary writes to prevent drift and incidents.

Ensure data model/DB designs cover constraints, integrity, indexes, migrations, rollback, and operational impacts.

Ensure architecture docs define RPO/RTO, failover paths, data consistency, and DR drillability.

Ensure event-driven designs specify delivery guarantees, ordering, idempotency, schema evolution, and replay/backfill strategy.

Ensure designs document third-party dependencies, SLAs, quotas, failure modes, and vendor lock-in mitigation.

Ensure designs specify failure modes, timeouts, error contracts, and observability for critical flows.

Detects over-permissive Firestore/Storage rules (allow read, write: if true / auth-only without ownership), missing auth checks on writes, and admin SDK private key exposure to client bundles.

Reviews GitHub Actions workflow diffs for script injection of untrusted input, pullrequesttarget with untrusted checkout, over-broad GITHUBTOKEN permissions, and unpinned third-party actions.