HarnessSkills and Packs
Skill Reference
Source-linked lookup for shared Harness skills
Skill reference is the practical lookup for the shared Harness skill library.
Use it when you need to know which skill exists, when it applies, which pack/category owns it, and where the executable SKILL.md source lives. The source of truth is the shared skills repository. Generated mirrors under .agents/skills and apps/cli/skills are distribution copies.
Direct source links in concept pages point readers to executable SKILL.md contracts when inspection helps; this catalog remains the full lookup.
Use Atomic Skills when a skill needs operator judgment beyond a lookup row.
Phases
| Skill | Use when | Pack/category | Source |
|---|---|---|---|
business-finder | Human-select optional Business intake and Product Area or Initiative projection for one Fog. | phases | SKILL.md |
functional-finder | Human-select optional Functional intake and Business structure or Epic projection. | phases | SKILL.md |
finder-phase | Run the shared human-only resumable Fog intake engine for an explicit Finder lens. | phases | SKILL.md |
requirements-phase | Close requirements, compile OUT-### outcomes, and project derived delivery identities. | phases | SKILL.md |
design-phase | Turn existing product evidence into confirmed design constraints and approved artifacts. | phases | SKILL.md |
delivery-phase | Resume scoped delivery by routing current artifacts to one lifecycle phase module at a time. | phases | SKILL.md |
review-phase | Review a diff, branch, PR, or artifact with findings-first evidence. | phases | SKILL.md |
debugging-phase | Investigate runtime failures with evidence, hypotheses, bounded fixes, and reruns. | phases | SKILL.md |
bug-discovery-phase | Run ClawPatch discovery and report generation without default source mutation. | phases | SKILL.md |
bug-resolution-phase | Classify and resolve selected ClawPatch findings through bounded Harness edits. | phases | SKILL.md |
docs-ingest-phase | Ingest proven changes into routed wiki/project docs and repo docs. | phases | SKILL.md |
resolve-debt-phase | Validate or resolve one canonical tech-debt artifact. | phases | SKILL.md |
goalify | Compile messy context into a compact /goal prompt. | phases | SKILL.md |
goalify lives in the phase tree for distribution, but it is a prompt compiler rather than a phase wrapper.
Agnostic Skills
| Skill | Use when | Pack/category | Source |
|---|---|---|---|
hi-cli | Operate hi, setup handoffs, updates, reports, and post-command artifacts. | agnostic/cli | SKILL.md |
architect-pipeline | Design or implement provider-neutral CI/CD, infrastructure deployment, release, and promotion workflows. | agnostic/security | SKILL.md |
debug-agent | Gather runtime evidence, logs, browser observations, or manual test support. | agnostic/debug | SKILL.md |
docs-onboarding | Reconstruct initial wiki/spec context from code, docs, and backlog context. | agnostic/docs | SKILL.md |
writing-fragments | Gather reusable raw material for substantial public-facing docs. | agnostic/docs | SKILL.md |
writing-beats | Order public-doc material into reader-facing beats before final shape. | agnostic/docs | SKILL.md |
writing-shape | Turn fragments and beats into a deliberate article or page structure. | agnostic/docs | SKILL.md |
writing-for-agents | Write skills, prompts, handoffs, and agent instructions with precise triggers. | agnostic/docs | SKILL.md |
handoff | Compact current session state into a clean continuation document. | agnostic/generic | SKILL.md |
create-spec | Compile closed decisions into an agent-ready provider-neutral SPEC.md. | agnostic/planning | SKILL.md |
create-plan | Turn a reviewed spec into a dependency-aware execution plan. | agnostic/planning | SKILL.md |
implement-spec | Execute an approved plan with notes, validation, and closeout evidence. | agnostic/planning | SKILL.md |
grilling | Run prerequisite-settled frontier rounds for requirements, plans, or design. | agnostic/requirements | SKILL.md |
domain-modeling | Sharpen domain language while a requirements grill is active. | agnostic/requirements | SKILL.md |
swarm-planner | Plan multi-agent execution only when explicitly invoked. | agnostic/subagents | SKILL.md |
requirements-grill | Pin requirements branches, glossary, axioms, parked scope, and wiki synthesis. | agnostic/requirements | SKILL.md |
write-backlog | Reconcile or materialize Linear/GitHub product topology, Finder projections, normalization, or observed delivery state. | agnostic/requirements | SKILL.md |
parallel-research | Fan out readonly research across independent questions or artifacts. | agnostic/research | SKILL.md |
prototype | Build one-file logic demos or switchable UI variants to answer a design question. | agnostic/research | SKILL.md |
wait-what | Request a context-restated, simplified-English re-pitch. | misc | SKILL.md |
improve-codebase-architecture | Surface architecture friction, poor boundaries, and module-depth opportunities. | agnostic/research | SKILL.md |
review | Frame reviews with separate Standards and Spec axes. | agnostic/quality | SKILL.md |
autoreview | Run the OpenClaw structured review helper as the core review runtime. | agnostic/quality | SKILL.md |
simplify | Reduce avoidable complexity, naming drift, and unnecessary abstraction. | agnostic/quality | SKILL.md |
tdd | Write or review changes through test-first or test-guided practice. | agnostic/quality | SKILL.md |
make-tsuite | Audit automated software test portfolios or apply authorized test-only changes without production changes. | agnostic/quality | SKILL.md |
backend-domain-structure | Organize backend code around domain boundaries. | agnostic/backend | SKILL.md |
backend-recoverable-actions | Design backend operations that can recover, retry, and report failures. | agnostic/backend | SKILL.md |
logging-best-practices | Improve structured logs, event shape, and observability context. | agnostic/backend | SKILL.md |
agent-browser | Verify local web flows with browser automation and evidence capture. | agnostic/frontend | SKILL.md |
frontend-domain-structure | Organize frontend application code around domain ownership. | agnostic/frontend | SKILL.md |
design-taste-frontend | Improve UI/UX quality, composition, visual hierarchy, and frontend taste. | agnostic/frontend | SKILL.md |
gpt-taste | Apply high-end UX/UI taste guidance for product surfaces. | agnostic/frontend | SKILL.md |
image-to-code | Generate website design images first, then implement the frontend to match them. | agnostic/frontend | SKILL.md |
imagegen-frontend-web | Generate one premium website design reference image per section. | agnostic/frontend | SKILL.md |
imagegen-frontend-mobile | Generate premium app-native mobile screen concepts inside phone mockups. | agnostic/frontend | SKILL.md |
redesign-existing-projects | Upgrade an existing web project without losing its product purpose. | agnostic/frontend | SKILL.md |
write-a-skill | Author a reusable Harness skill. | agnostic | SKILL.md |
grilling preserves Matt Pocock's MIT-licensed batch-grill-me primitive pinned at 9603c1cc8118d08bc1b3bf34cf714f62178dea3b. Devpunks adapts invocation metadata while requirements, planning, onboarding, and design wrappers compose the primitive.
domain-modeling is imported from its pinned upstream source, which remains the authority for modeling behavior. requirements-grill invokes it with the active grill glossary as working persistence. After accepted closure, the existing wiki-synthesis step promotes accepted terminology into the routed glossary for the relevant bounded context.
wait-what is user-invoked. It asks for the same explanation with restored context, ASD-STE100 Simplified Technical English, and stable vocabulary from GLOSSARY.md; it is not an automatic prose filter.
architect-pipeline is part of the default security pack. It owns provider-neutral CI/CD and infrastructure-deployment orchestration, including release identity, artifact creation, publication, retry, and promotion theory. Route explicit pipeline security audits to audit-cicd-security; route application and language security work to security-best-practices. The Pulumi pack owns Pulumi APIs and resource, state, registry, authentication, ESC, and Automation API specifics.
Framework Skills
| Skill | Use when | Pack/category | Source |
|---|---|---|---|
better-auth-best-practices | Integrate or review Better Auth behavior. | frameworks/better-auth | SKILL.md |
effect | Build production TypeScript applications with Effect v4. | frameworks/effect | SKILL.md |
effect-backend-structure | Structure Effect backend modules and ownership. | frameworks/effect | SKILL.md |
effect-recoverable-actions | Apply recoverable-action patterns in Effect code. | frameworks/effect | SKILL.md |
elysiajs | Build or review ElysiaJS backend code. | frameworks/elysia | SKILL.md |
nestjs-best-practices | Review or implement NestJS best practices. | frameworks/nestjs | SKILL.md |
nestjs-expert | Build or debug NestJS systems with deeper framework guidance. | frameworks/nestjs | SKILL.md |
next-best-practices | Build or review Next.js App Router behavior. | frameworks/nextjs | SKILL.md |
next-cache-components | Work with Next.js Cache Components. | frameworks/nextjs | SKILL.md |
pulumi-overview | Route Pulumi IaC, Pulumi Cloud, and related Pulumi work. | frameworks/pulumi | SKILL.md |
pulumi-best-practices | Write or review reliable Pulumi programs. | frameworks/pulumi | SKILL.md |
pulumi-component | Author Pulumi ComponentResource classes. | frameworks/pulumi | SKILL.md |
pulumi-automation-api | Use Pulumi Automation API safely. | frameworks/pulumi | SKILL.md |
pulumi-esc | Work with Pulumi ESC environments, secrets, and configuration. | frameworks/pulumi | SKILL.md |
provider-upgrade | Upgrade Pulumi providers without unintended infrastructure changes. | frameworks/pulumi | SKILL.md |
package-usage | Track package usage and versions across Pulumi organization stacks. | frameworks/pulumi | SKILL.md |
async-react-patterns | Apply React async UI and data-flow patterns. | frameworks/react | SKILL.md |
vercel-composition-patterns | Review React composition and component architecture. | frameworks/react | SKILL.md |
vercel-react-best-practices | Apply React performance and correctness guidance. | frameworks/react | SKILL.md |
tanstack-query | Manage server state with TanStack Query. | frameworks/tanstack-query | SKILL.md |
trpc/tanstack-query | Apply TanStack Query patterns in a tRPC surface. | frameworks/trpc | SKILL.md |
turborepo | Work with Turborepo structure, tasks, cache, filtering, and CI. | frameworks/turborepo | SKILL.md |
Language Skills
| Skill | Use when | Pack/category | Source |
|---|---|---|---|
quality-types | Improve TypeScript type quality and boundary safety. | languages/typescript | SKILL.md |
async-python-patterns | Apply async Python design and implementation patterns. | languages/python | SKILL.md |
python-code-style | Improve Python readability, conventions, and style. | languages/python | SKILL.md |
python-design-patterns | Apply Python design patterns without overcomplication. | languages/python | SKILL.md |
python-backend-structure | Compound backend-domain-structure with Python packages, import direction, public APIs, and composition roots. | languages/python | SKILL.md |
python-testing-patterns | Write or review Python tests and test architecture. | languages/python | SKILL.md |
python-project-structure is retained only as a deprecated compatibility ID that redirects to python-backend-structure.