Skill Claude Code
Accessibility auditor mapping a UI surface (screen, flow, or component set) to WCAG 2.2 at a named conformance level, as a per-criterion conformance ledger. Activates when a screen-spec or implemented surface has no ledger, or before sign-off. Do not use for contrast-only requests (use color-contrast-architect for 3+…
Skill Claude Code
Design an eval plan for a product AI feature: success criteria, a held-out labeled dataset, per-criterion grading (code-based, then LLM-based), and a pass threshold. Use when a task ships or changes a feature with non-deterministic model output (classification, extraction, summarization, ranking, agent action) needing…
Skill Claude Code
Review an API contract (shapes, errors, auth model) BEFORE implementation for naming, versioning, pagination, idempotency, and fit with existing endpoints. Use when designing or changing endpoints. Do not use when the API is already implemented (use review-hard or repo-consistency-sweep).
Skill Claude Code
Verify a built backend HTTP surface at runtime: per route, record the response, assert it against the slice's acceptance behavior, and decide PASS/FAIL/BLOCKED. Do not use to review a contract before implementation (use api-contract-review or graphql-contract-review), write code (use implement-approved-slice), triage…
Skill Claude Code
Verify a mobile or app runtime: run the app (device, emulator, or headless), read captured output (logcat, iOS device log, or Metro/JS console), classify errors per stack, and decide a PASS/FAIL gate. Adapters: React Native/Expo, Unity mobile. Do not use to plan a screen (use implementation-plan), to write or fix code…
Skill Claude Code
Persist triage decisions (apply, decline, discuss) into REVIEWPREFERENCES.md so future sweeps suppress declined findings. Use after editing triage values in a SWEEP snapshot from repo-consistency-sweep. Do not use without a snapshot or before triage values are set.
Skill Claude Code
Atomically lock IMPLEMENTATIONPLAN.md as the approved execution baseline and stamp TASKSTATE.md with the approval signal; the plan-level counterpart to approve-proposed. Use on every plan implementation-plan produces or self-critique-and-revise revises; it self-runs a blinded review, and only an ESCALATED exit reaches…
Skill Claude Code
Atomically persist every file marked PROPOSED in the most recent prior assistant turn's ### Artifact changes block, promoting staged content to real files. Invoked on request only. Use when that block contained files marked PROPOSED and you have accepted the inline content of each. Do not use when the prior turn had…
Skill Claude Code
Audit every atom in parallel against the component guidelines, one worker per small group, merging rows into one ATOMAUDIT.md table (orchestrator-workers variant of atom-audit). Use when the atom count is 6 or more and COMPONENTGUIDELINES.md exists. Do not use below 6 atoms (use atom-audit single-agent), without…
Skill Claude Code
Produce ATOMAUDIT.md table auditing every atom component against COMPONENTGUIDELINES.md (memo, callbacks, inline styles, touch target, a11y, reduced motion). Use when 2-4 weeks have passed since last audit, 5+ new atoms shipped, or COMPONENTGUIDELINES has a new normative rule. Do not use when no atoms exist (run…
Skill Claude Code
Read-only board of an autonomous-run task: IMPLEMENTATIONPLAN waves and slices and TASKSTATE phases as to-do, in-progress, escalated, proposed, done columns. Use when you want a single-glance status of an autonomous run. Do not use to change state (use sync-task-state), to assess a multi-slice task (use where-we-at)…
Skill Claude Code
Decide whether an already-defined project is ready to boot an unattended overnight run, and refuse until it is. Emits a per-criterion ledger with a BOOT or NOT-READY verdict naming every missing item. Use before handing a night to autonomous-run. Do not use to approve a plan (use approve-plan), to drive the run itself…
Skill Claude Code
Run one approved task in one supervised session with the direct-use reference dispatcher, delegating verifiable slices to existing commands and emitting PROPOSED diffs only and never creating commits or attestation refs. Use when a maintainer invokes Fhorja directly after approve-plan and a BOOT verdict from…
Skill Claude Code
Produce a staff-grade backend system-design RFC: a 12-section document for a new service, endpoint, or feature. Composes with slo-define, performance-budget, api-contract-review, release-plan. Use when a backend feature needs architecture-level design before planning. Do not use to slice a designed change (use…
Skill Claude Code
Return a branch name and concise commit message (max 3 lines) for the current task, grounded in the real git diff; with --apply, create the commit after showing the staged diff. Use when the user only needs quick branch and commit naming before committing, PR packaging is unnecessary, and a real diff exists; use…
Skill Claude Code
Append one observation, question, hypothesis, or concern to TASKSTATE.md as lean, append-only task memory, leaving other artifacts and work undisturbed. Use when something surfaces mid-work, a hypothesis needs logging, or a concern is too small for decision-interview or targeted-questions. Do not use for a decision…
Skill Claude Code
Pull references (URLs or topics) from the web, summarize each with a freshness format, and append them to the project's REFERENCES.md for later tasks. Use when the user persists project-level references, supplies URLs, has a reference that outlives the task, or seeds REFERENCES.md. Do not use when the project folder…
Skill Claude Code
Generate and re-sync an AI-readable code context map (ranked, token-budgeted Markdown of files, imports, signatures, and typed boundary calls) for a target project (or, from a seed file, its import chain). Opt-in; not part of default task init. Use when fast structural orientation is needed before editing, when…
Skill Claude Code
Locate code paths and line ranges for a described behavior: up to 10 candidates with HIGH/MEDIUM/LOW confidence and a search trail, plus a proposed SOURCEOFTRUTH.md update. Use when the first question is where one specific behavior lives and paths are unknown. Do not use when paths are explicit, the codebase is tiny…
Skill Claude Code
Audit contrast pairs against WCAG 2.2 AA and AAA before visuals lock, with a matrix and token fixes. Activates when design-bootstrap proposes tokens without contrast pairs, or screen-spec or component-spec has an unvalidated pair. Do not use for a single pair (use component-spec), with no accessibility target (use…
Skill Claude Code
Produce a lossy compaction of TASKSTATE.md when task memory outgrows a useful size, keeping canonical decisions and the next step, dropping stale facts. Use when memory spans 5+ completed slices and feels heavy, resume cost grows, known facts are mostly resolved, or before a session pause. Do not use in early…
Skill Claude Code
Generate a 15-section component specification from a Figma component using MCP tools, matching the COMPONENTSPEC.md template and pre-filled from Figma. Use when documenting a design system component. Do not use when the component has no Figma representation or the spec already exists (edit manually instead).
Skill Claude Code
Harden the current decision set into a normative source of truth with no residual ambiguity, saved in DECISIONS.md and TASKSTATE.md as reviewable edits. Use when decisions are resolved, the contract direction is stable, and wording needs hardening before planning or implementation. Do not use when major ambiguities…
Skill Claude Code
Validate that a non-Supabase Postgres database (Cloud SQL, RDS, self-hosted) is reachable, introspect a scoped subset of its schema, and persist it as DBCONTEXT.md. Read-only, opt-in. Use when work touches that database and planning needs verified schema, or DBCONTEXT.md is stale. Do not use without a task (run…
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: