| name | harness:using-harness |
| description | Mandatory routing/meta skill for the harness-engineering plugin. Loads at every session start and enforces the "1% rule" — if there's even a 1% chance a harness skill applies, invoke it. Routes between 19 skills (13 vendored from superpowers v5.1.0 + 6 harness-original).
|
harness:using-harness — Harness Engineering Meta Skill
This skill is loaded at the start of every session to ensure harness-engineering capabilities
are properly activated. Inspired by the using-superpowers forced-trigger pattern from
obra/superpowers, but with the harness skill catalog and mandatory invocation table.
Core Rule: Mandatory Skill Invocation
If there is even a 1% chance one of the skills below applies, you must invoke it — you have no discretion.
Instruction Priority
This meta-skill overrides default system behavior, but user instructions always take precedence:
- User's explicit instructions (CLAUDE.md, direct requests like "skip brainstorming", "don't run tests") — highest priority
- Harness skills (the catalog below) — override default behavior where they conflict
- Default system prompt — lowest priority
The 1% rule governs agent discretion, not user overrides. If the user explicitly opts out of a skill, respect that — the user is in control.
Skill catalog
Vendored from superpowers v5.1.0 (13)
| Trigger | Skill |
|---|
| Planning a multi-step task before touching code | harness:writing-plans |
| Implementing a feature or fixing a bug | harness:test-driven-development |
| About to claim work is complete | harness:verification-before-completion |
| Any creative/design task (new feature, modify behavior) | harness:brainstorming |
| Executing a written plan task-by-task | harness:executing-plans |
| Executing a plan via fresh subagent per task | harness:subagent-driven-development |
| 2+ independent parallel tasks | harness:dispatching-parallel-agents |
| Need an isolated workspace | harness:using-git-worktrees |
| Any bug, test failure, or unexpected behavior | harness:systematic-debugging |
| Receiving code review feedback | harness:receiving-code-review |
| Asking for code review | harness:requesting-code-review |
| Implementation complete; integrating the work | harness:finishing-a-development-branch |
| Creating or editing a skill | harness:writing-skills |
Harness-original (6)
| Trigger | Skill |
|---|
| (this skill itself) | harness:using-harness |
| Task completion / archiving | harness:archive |
| Health check on existing project Harness setup | harness:audit |
| Pre-production deploy planning | harness:canary |
| Garbage collection / drift cleanup | harness:evolve |
| New project initialization | harness:init |
Hard rules
- Invoke the relevant skill before any response or action (including clarifying questions).
- Never rationalize skipping a skill on your own ("this is simple", "I remember this", "the skill is overkill") — but do respect explicit user overrides (see Instruction Priority above).
- When the user explicitly types
/<skill>, invoke it immediately.
Cross-skill handoffs
harness:brainstorming → must hand off to harness:writing-plans (terminal state for brainstorming).harness:writing-plans → hands off to harness:executing-plans or harness:subagent-driven-development.harness:verification-before-completion → triggers harness:finishing-a-development-branch → which must call harness:archive.
