Spec-driven AI development workflow — methodology docs, starter templates, and Claude Code/Codex plugins with feature specs, current-truth docs, ADRs, hooks, and review gates.
Agent
Claude Code adapter for L2 project-convention compliance review. Reads docs/reviewers/agents-md-reviewer.md as the canonical reviewer spec, then reviews changed code against AGENTS.md and path-scoped rules with complete-population evidence. Does not review functional correctness or design taste.
Agent
Claude Code adapter for existing-codebase structure survey. Reads docs/reviewers/codebase-explorer.md as the canonical explorer spec, then reports directories, tiers, frameworks, modules, and an AGENTS.md Project Structure snippet. Read-only.
Agent
Claude Code adapter for generated-decision traceability audit. Reads docs/reviewers/decision-completeness-auditor.md as the canonical auditor spec, then checks generated content for unanchored specific-string decisions before preview/apply gates. Read-only.
Agent
Claude Code adapter for pre-implementation Requirements Reconciliation and Q3-Q7 artifact quality, including bounded architecture adequacy inside Q5/Q7c/Q7d when the accepted delivery shape has a real architecture boundary signal. Reads docs/reviewers/spec-quality-reviewer.md and does not review code-vs-spec…
Agent
Claude Code adapter for L3 implementation-vs-feature compliance. Reads docs/reviewers/spec-reviewer.md; checks the accepted feature record and relevant optional attachments. Domain docs are context only.
Agent
Claude Code adapter for technical choice research. Reads docs/reviewers/tech-researcher.md as the canonical researcher spec, then compares a focused candidate set and recommends one default with rationale. Read-only.