Checks transaction scope, missing rollback handling, long-held transactions, trigger/notify interaction. Returns findings with severity, location, effort.

Research best practices via MCP Ref/Context7/WebSearch and create documentation (guide/manual/ADR/research). Single research, multiple output types.

Builds dependency graph, detects cycles (DFS), validates boundary rules, calculates coupling metrics (Ca/Ce/I, CCD/NCCD). Adaptive architecture detection.

Creates Stories from IDEAL plan (CREATE) or appends user-requested Stories (ADD). Generates 9-section documents, validates INVEST, creates in Linear.

Creates task management documentation (docs/tasks/README.md + kanban_board.md). Sets up Linear integration and task tracking rules.

Generates GitHub Actions CI workflow configuration

Checks outdated packages, unused deps, reinvented wheels, vulnerability scan (CVE/CVSS). Supports mode: full | vulnerabilities_only.

Coordinates project documentation creation. Gathers context once, detects project type, delegates to specialized workers (ln-111-115).

Configures Swagger/OpenAPI documentation

Traffic-First opportunity discovery. KILL funnel filters ideas by traffic channel, demand, competition, revenue, interest, MVP-ability. Outputs one idea + one channel recommendation.

Checks async races, thread safety, TOCTOU, deadlocks, blocking I/O, resource contention, cross-process races. Two-layer detection: grep + agent reasoning.

Checks unreachable code, unused imports/variables/functions, commented-out code, backward compatibility shims, deprecated patterns. Returns findings.

Checks bootstrap initialization order, graceful shutdown, resource cleanup, signal handling, liveness/readiness probes. Returns findings.

Orchestrates final verification - build, test, and container health checks

Executes all test suites and reports results with coverage

Performs manual testing of Story AC via executable bash scripts saved to tests/manual/. Creates reusable test suites per Story.

Checks hardcoded secrets, SQL injection, XSS, insecure dependencies, missing input validation. Returns findings with severity, location, effort, recommendations.

Audits layer boundaries + cross-layer consistency: I/O violations, transaction boundaries (commit ownership), session ownership (DI vs local).

Worker that runs existing tests to catch regressions. Auto-detects framework, reports pass/fail. No status changes or task creation.

Coordinates code quality: ln-511 metrics, ln-512 cleanup, inline agent review (Codex + Gemini), ln-513 regression, ln-514 log analysis. Returns quality_verdict + aggregated results.