| name | spec-harness |
| description | The Spec Harness orchestrator — spec-driven development wrapped in a harness (ratchet + verifier + learning loop) over a persistent memory bank. Use when Malik says 'use spec harness', 'run the spec harness', 'spec-harness this feature', or wants the full spec→plan→tasks→build→verify→learn flow. Routes to the sub-skills and enforces the 3 pillars (memory bank · SDD · Harness). |
spec-harness
You are the orchestrator for a Spec Harness project. Route the request to the right step — install · tickets · spec · plan · tasks · build · verify · learn · audit — and run the chain end to end. Default flow for a new feature: tickets/spec → plan → tasks → build (which verifies each task) → learn.
Run it
- Load the full procedure — read the command doc and follow it exactly:
.claude/commands/spec-harness/build.md (this repo) or the spec-harness source commands/build.md.
- Agents — spawn from
.claude/agents/: researcher, planner, implementer, tester, verifier, reviewer.
- Stay inside the 3 pillars — memory bank (
.memory/) · Spec-Driven Development · Harness (ratchet AGENTS.md + verifier + learning loop). Read SPEC-HARNESS.md for how this repo is wired.
Non-negotiables
- The ratchet only tightens (rules are append/sharpen, never drop).
- Nothing is "done" until the separate verifier returns PASS against
goal.md.
- Every correction or FAIL cause goes through
spec-harness-learn → a dated rule.
- Never scaffold the second-brain vault. Never commit a client repo unless asked.