This repository also configures its own agents. See what skills tells them →
Agent Claude Code
Designs one decisive implementation blueprint for a feature - vertical slices with frozen per-slice API contracts, change-site map, data changes, dependency graph - and writes the six plan documents directly into the user-story folder. Returns a short manifest, not options.
Agent Claude Code
Traces how an existing feature or domain area is implemented - entry points, execution flow, conventions, reusable code, key files. Dispatched in parallel by the orchestrator in Phase 2.
Agent Claude Code
Implements the backend half of one vertical slice via TDD against its ACs and frozen contract (e.g. "SLICE-01 backend half"), or bootstraps a new backend (scope "project scaffold"). Never touches other slices.
Agent Claude Code
Implements the frontend half of one vertical slice via TDD against a typed mock of its frozen contract (scope "SLICE-01 frontend half"), reconciles every slice's mock against the real backends once (scope "whole-story integration"), or bootstraps a new frontend (scope "project scaffold"). Never touches other slices…
Agent Claude Code
Simplifies the whole-story diff once, in an isolated context, before the reviewers see it, conforms it to the house conventions (react-best-practices, restful-api-design) without changing behaviour, runs the tests covering the files it touched, commits, and returns a Return Report.
Agent Claude Code
Reviews the six plan documents (not code) for AC coverage, slice integrity, real-vs-imagined dependencies, security, test coverage and cross-document consistency. Dispatched in parallel pairs by the orchestrator in Phase 6.
Agent Claude Code
Runs the Phase 9 walkthrough in a clean context: writes a Playwright spec per slice from 05-test-plan.md that self-captures screenshots, runs them headless, writes 06-walkthrough.md, persists the specs into the project's existing e2e suite, and returns a Return Report. Reports app bugs; never patches app code.
Agent Claude Code
Reviews the story's implementation diff against the slice cards' ACs and the technical plan: AC intent vs test assertions, error paths, convention adherence. Runs in parallel with security-reviewer. Reports a verdict; never patches code.
Agent Claude Code
Security review of the story's implementation diff against the technical plan's threat surface: authorisation, input validation, injection, secrets/PII exposure, auth-adjacent controls, dependencies. Runs in parallel with code-reviewer. Reports a verdict; never patches code.
Agent Claude Code
Builds, verifies and lands a new project's walking skeleton from the skeleton contract and stack rows that project-setup hands it: repository and worktree, both halves in parallel through impl-backend and impl-frontend, an evidence check and an end-to-end smoke, project context and README, the merge onto the base…
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: