Skill index

Every skill, alphabetical, with a one-line purpose and its phase page. The runtime serves 71 skills. Open a phase page for the deep per-skill entry (what it does, when to use it, inputs, outputs, hard gates, and handoffs).

For the skill-to-skill flow, see The skill workflow.

Skill	Purpose	Phase page
align-grid	Build an editorial webpage on a verified Müller-Brockmann grid	Build
assess-impact	Analyze the blast radius of a change before code	Plan
audit-code	Self-review checklist before dispatching a reviewer	Verify
audit-plan	Evaluate an incoming plan, produce a READY verdict	Utility
change-request	Add a requirement or reorder work by selected project policy	Plan
commit-message	Draft a Conventional Commits message and its SemVer bump	Release
compose-workflow	Chain multiple skills into a custom workflow recipe	Sustain
context7-mcp	Fetch current library docs through the Context7 server	Utility
craft-skill	Create a new skill with the correct structure	Build
deepen-architecture	Find deepening opportunities in a codebase	Design
define-language	Extract a ubiquitous-language glossary from the conversation	Design
delegate-task	Delegate one complex task to a subagent with two-stage review	Sustain
deploy	Build, verify, deploy, wait, then smoke the deployment	Build
design-interface	Generate several interface designs, then compare	Design
develop-tdd	Test-driven development with a red-green-refactor loop	Build
diagnose-stall	Diagnose why agent orchestration stopped	Utility
dispatch-agents	Dispatch multiple subagents in parallel	Sustain
elaborate-spec	Refine a rough idea into a clear specification	Discover
enforce-first	Apply the F.I.R.S.T test-quality rubric	Verify
evolve-skill	Benchmark-gated skill evolution	Sustain
execute-group	Execute active work with checkpoint or autonomous control	Build
extract-design	Extract a DESIGN.md from an HTML prototype	Utility
find-way	Map a large effort as decision tickets on a tracker	Utility
fix-bug	Orchestrate the bug-fix chain	Verify
grill-me	Challenge a plan in context-only or docs mode	Design
guard-git	Block a dangerous git command, enforce opt-in policy	Build
hook-commits	Set up pre-commit hooks for Node.js projects	Build
inspect-quality	An interactive QA session that logs bugs to the registry	Verify
investigate-bug	Investigate a bug, find the root cause, write a fix plan	Verify
kickoff-branch	Create an isolated worktree and verify a clean baseline	Build
maintain-wiki	Keep the concept-wiki bundle consistent with its sources	Utility
map-codebase	Derive the tech-stack note by scanning the codebase	Discover
migrate-spec	Transform a foreign spec artifact into the project layout	Sustain
model-domain	Stress-test a plan against the domain model, capture invariants	Design
orchestrate-project	Coordinate a multi-phase project with hard gates	Build
organize-workspace	Scan for disposable artifacts, propose a safe tidy	Sustain
plan-refactor	Create a refactor plan of tiny commits	Plan
plan-release	Sequence tasks and optional groups by project policy	Plan
plan-tests	Design a risk-scaled test architecture for a group	Plan
plan-work	Write detailed, verifiable implementation tasks	Plan
publish-package	Publish a package to npm, crates.io, PyPI, or Homebrew	Build
quick-fix	A fast path for a trivial data-only fix	Build
release-branch	Verify the gates and land a finished branch	Release
request-review	Dispatch risk-scaled independent reviewers	Verify
research-first	Search for prior art before implementing	Discover
reset-baseline	Restore the project to a known clean state	Sustain
respond-review	Act on reviewer feedback systematically	Verify
run-benchmark	Run isolated skill benchmarks with validation deltas	Utility
run-evals	Eval-driven development, define evals before building	Verify
scope-work	Define what is in and out of scope (spine step 1)	Plan
security-review	Security analysis of code changes across files	Verify
seed-conventions	Generate the agent guide and conventions for a new project	Plan
session-state	Track decisions and progress to prevent context rot	Sustain
setup-environment	Pre-install dependencies and configure tools	Build
simple-english	Write technical text with Simplified Technical English	Utility
slice-tasks	Break a scoped PRD into vertical slices (spine step 2)	Plan
smoke-test	Post-deploy health check against a live URL	Build
spike-prototype	A throw-away prototype for an unknown problem space	Build
stocktake-skills	A batch audit of the skill catalog	Sustain
survey-context	Read the current state, map the phase, suggest the next skill	Discover
trace-requirement	Report and gate task-to-evidence traceability	Verify
using-truenorth	The one-time bootstrap and routing entry point	Discover
validate-contracts	Assert data-shape consistency across boundaries	Build
validate-fix	Prove a fix works and harden against recurrence	Verify
verify-work	The multi-phase UAT gate	Verify
visual-dashboard	A read-only browser dashboard of the project status	Utility
wire-ci	Set up GitHub Actions CI with local validation	Build
wire-observability	Add logging, observability, and setup scripts scaled to risk	Build
write-document	Create or revise a high-integrity technical document	Sustain