Skill Claude Code
Validate that a Supabase MCP server is reachable, introspect a subset of the database (tables, columns, RLS policies), and persist it as DBCONTEXT.md. Read-only, never destructive SQL. Falls back to the local Supabase CLI when no MCP server is reachable. Use when the task touches Supabase data, schema or RLS and…
Skill Claude Code
Ask the minimum decision questions needed before turning the task into canonical implementation rules, on behavior, data integrity, rollout safety, or test strategy. Use when missing information is decision-driven (not factual) or undecided policy blocks planning; in an attended chain it records an open decision as…
Skill Claude Code
Generate an outward-facing delivery artifact (executive summary, slack or email update, demo script, release note, blog post draft) per audience and format, grounded in task artifacts with no invented claims. Use when delivery needs audience-specific framing beyond a PR description or team status note. Do not use for…
Skill Claude Code
Bootstrap a design system from a Figma file via MCP: extract token variables, identify components, scaffold foundation docs and a component inventory. Use when starting a new project with a Figma file. Do not use when docs exist or there is no Figma file.
Skill Claude Code
Review a component or screen against its spec (variants, states, accessibility, tokens). Distinct from review-hard (general risk) and repo-consistency-sweep (pattern matching). Activates when a slice's Scope touches a design-system component with no cited design-spec-review pass. Use when an implemented component or…
Skill Claude Code
Capture a small-to-medium course correction found mid-task in your own work: record a D-N entry in DECISIONS.md, update TASKSTATE.md, and route back to the command. Use when it is worth recording, does not invalidate the whole approach, and the current slice stays recoverable with a small plan change. Do not use for…
Skill Claude Code
Orchestrator-workers variant of external-research: parallel workers each synthesize one angle of a research question, merged into one EXTERNALRESEARCH.md. Use when the research question has N >= 3 distinct angles or source-groups AND would benefit from parallel deep reads. Do not use when 1-3 sources suffice (use…
Skill Claude Code needs its repo
Synthesize multiple external sources into a task-scoped EXTERNALRESEARCH.md grounded in REFERENCES.md, with comparative analysis and a recommendation. Use when sources need synthesis, not capture. Do not use for one source or project research (use capture-references), internal questions (use code-locate or…
Skill Claude Code
Extract canonical foundations docs (color, typography, spacing, radii) from a batch of existing SCREENSPECs. Re-runs only add new tokens and route conflicts to a Review queue. Use when SCREENSPECs already document raw values and need converging into role tokens. Do not use when foundations are already curated and…
Skill Claude Code
Orchestrator-workers variant of feature-library-scout for deep per-feature-problem library research. One worker per problem ranks libraries by adoption signal. Use when over 3 distinct feature problems each need a deep read. Do not use for 1 to 3 problems (use feature-library-scout), stack layers (use…
Skill Claude Code
Research and recommend vetted libraries per feature problem (lists, camera, forms, keyboard, sheets), ranked by adoption signal, writing FEATURELIBRARIES.md from captured REFERENCES.md sources. Use when the stack is chosen and you want per-feature libraries vetted. Do not use to pick stack layers (use…
Skill Claude Code
Compare design tokens in code against foundation docs and optionally Figma variables to detect drift (undocumented tokens, tokens missing in code, value mismatches). Use when the design system has evolved and foundations need a sync check. Do not use when no foundation docs exist (run design-bootstrap first).
Skill Claude Code
Review a frontend architecture at scale, gating micro-frontend adoption BEFORE building (default: modular monolith until 3 or more independently deploying teams and real coordination pain exist). Do not use to design one system (use frontend-system-design), to review a GraphQL or REST contract (use…
Skill Claude Code
Produce a frontend system-design RFC for the active task (12 sections, web and mobile) as FRONTENDSYSTEMDESIGN.md, with an --interview mode for interview rounds. Do not use to frame the problem (use problem-framing), to slice a designed change (use implementation-plan), for blast radius (use impact-analysis), for an…
Skill Claude Code
Verify a built Godot 2D or 3D scene at runtime: run it, read the debugger output, classify errors against a Godot taxonomy, decide a PASS/FAIL gate. Real output is the evidence. Use after a Godot slice is implemented, to gate runtime behavior. Do not use to plan a scene (use godot-scene-plan), write or fix code (use…
Skill Claude Code
Plan the Godot scene and node structure for a 2D or 3D game feature before GDScript is written: scene tree, nodes, autoloads, signals, input map. A 3D plan SHALL declare its renderer tier. Do not use for game-idea framing (use problem-framing in its game-design mode), slicing a plan (use implementation-plan), blast…
Skill Claude Code
Review a GraphQL schema and a Backend-for-Frontend (BFF) contract BEFORE implementation against a GraphQL checklist (nullability, N+1, cost limits, federation, BFF thinness). Use when designing or modifying a GraphQL schema, a federated subgraph, or a BFF contract. Do not use for a REST contract (use…
Skill Claude Code
Scan the session and task artifacts for reusable lessons (what was tried, what failed and why, what surprised us) and propose entries to append to LEARNINGS.md, the produce-side counterpart to what task-init reads. Use on demand mid-task after a hard-won fix or surprising failure, or at closure to sweep a long…
Skill Claude Code
Break the task out of a loop, confusion state, or false progress and find the fastest safe path forward, routing to a recovery command and editor mode. Use when progress loops or stalls, questions repeat, the wrong command or mode may be in use, or the user is unsure if it is technical, workflow, or scope. Do not use…
Skill Claude Code
Generate a design-system spec from an image (screenshot, mockup, or app screen) when there is no Figma source: a component spec (--component) or screen spec (--screen), auto-detected. A gated --gameplay mode (off by default) derives a game MECHANICSSPEC.md from screenshots and playtest notes. Use when you have only an…
Skill Claude Code
Understand a requested change well enough to make safe workflow decisions, and persist IMPACTANALYSIS.md: blast radius, contract, schema and runtime risks. Use when a task is new or partly understood, or a change may touch contracts, schema, integrations, runtime, or critical user flows. Do not use if a valid one…
Skill Claude Code
Implement only the approved slice with minimal changes, then persist execution evidence in slice notes and TASKSTATE.md. The official execution path. Opt-in test-first mode via --tdd. Use when the plan and slice are approved, ambiguity is resolved and files in scope are known. Do not use during discovery or planning…
Skill Claude Code
Orchestrator-workers variant of implement-approved-slice that runs independent approved slices in parallel, one worktree-isolated worker per slice per wave, merged after each wave. Use when the active task has an approved multi-slice plan whose Execution waves show at least one wave of size 2 or more. Do not use when…
Skill Claude Code
Execute a bounded micro-delta (fixes, polish, missed checklist items) inside the same slice intent and locked decisions, with evidence in slice notes. Use when a slice is implemented or ready-to-close-with-follow-ups and a narrow gap (bug, typo, test, log line, copy, refactor) fits a numbered micro-delta list over…
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: