From harness
One-sentence description of when to invoke this skill. Should answer 'what does this do' and 'when do I use it' in a single sentence. Used by the harness to surface the skill in the directory.
How this skill is triggered — by the user, by Claude, or both
Slash command
/harness:_templateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> Canonical skill template. Copy this directory to `skills/<your-skill>/` and replace placeholders. The `/harness:harness-audit` step `skill-structure-drift` validates every skill against this layout.
Canonical skill template. Copy this directory to
skills/<your-skill>/and replace placeholders. The/harness:harness-auditstepskill-structure-driftvalidates every skill against this layout.
When this skill should fire. Be specific — if a sibling skill already covers part of the territory, name it and explain the boundary.
/sibling-skill).What this skill expects to be available before it starts. Include file paths the skill will read, environment vars it reads, prior phase verdicts it depends on.
$state_dir/{task-id}/build.md with verdict BUILD_COMPLETECLAUDE.md Commands section, .env for secretsNumbered, auditable steps. One concrete action per step. The skill body MUST be procedural — readers execute it literally, not philosophically.
Concrete actions. Show the exact commands or tool invocations. If the skill spawns an agent, show the spawn invocation.
# Example concrete command
git -C "$WORKTREE" diff main...HEAD
...
If this step has a gate, name the gate and the verdict it produces.
What this skill produces. Be specific about file paths, frontmatter, and downstream consumers.
$state_dir/{task-id}/<phase>.md with the verdict in frontmatter$state_dir/{task-id}/scratchpad/<role>-<phase>.md for findings the next phase needs---
task_id: {task-id}
phase: <this-skill's-phase>
verdict: <one of the verdicts below>
timestamp: ISO-8601
---
## Summary
1-3 sentence outcome.
## Key Findings
- ...
## Next Phase Input
What the consuming phase needs to know.
The exact set of verdicts this skill emits. MUST match protocols/verdict-catalog.md. Use uppercase snake-case identifiers.
| Verdict | Meaning | Downstream |
|---|---|---|
VERDICT_NAME | Success path. | Next phase consumes the state file. |
VERDICT_NAME_FAILED | Failure path with concrete remediation. | Pipeline halts or returns to prior phase. |
The skill MUST emit exactly one verdict per invocation. Multiple verdicts on one run is a structural bug.
What NOT to do in this skill. Pulled from real incidents — keep this list short and specific to this skill's territory.
Skill-level tests live in skills/<your-skill>/tests/. At minimum:
protocols/verdict-catalog.md, and every section heading from this template exists.See skills/_template/tests/ for placeholder fixtures.
npx claudepluginhub paulingham/.claude --plugin harnessProvides UI/UX resources: 50+ styles, color palettes, font pairings, guidelines, charts for web/mobile across React, Next.js, Vue, Svelte, Tailwind, React Native, Flutter. Aids planning, building, reviewing interfaces.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.