Harness Help — Command Index & Scenario Quick Reference
--- description: Triggered when the user says "harness help", "harness commands", "harness usage", "what commands are available", or "show harness commands". Displays the Harness command index, installed Skill inventory, and high-frequency scenario entry points. ---
SKILL.md
Full skill instructions
Harness Help — Command Index & Scenario Quick Reference
description: Triggered when the user says "harness help", "harness commands", "harness usage", "what commands are available", or "show harness commands". Displays the Harness command index, installed Skill inventory, and high-frequency scenario entry points.
Behavior
- Display Harness command index
- Scan
~/.claude/skills/to list installed Skills by category - Show high-frequency scenario → command/Skill quick entry points
Command Index
Harness Command System:
harness — Project initialization (full 8-step workflow)
harness help — This help page (command index + scenario quick ref)
harness audit — Project health check (verify CLAUDE.md/docs/hooks/skill completeness)
harness quality gate — Pre-commit quality gate (tests + lint + security review + doc sync)
harness guide — Skill recommendation (match the best Skill for your scenario)
harness cleanup — Interactive temp file archive (scan → confirm → move to archive/, never deletes)
harness resume — Lightweight context recovery (same session after /compact, ~3k tokens)
harness handoff — Deep context handoff (new agent / crash recovery, ~8k tokens)
Installed Skills Inventory
Scan ~/.claude/skills/ directory, display by category:
🔧 Infrastructure (behavior control + planning + knowledge extraction):
[status] superpowers — brainstorming/TDD/debugging/code-review/verification
[status] planning-with-files — /plan task planning + 4 hooks for continuous injection
[status] claudeception — lessons learned → auto-generate Skills
🏭 Skill Factories (generate new Skills):
[status] skill-creator — codify workflows into Skills
[status] security-review-skill-creator — generate security audit rules for a project
🔒 Security:
[status] security-review-skill-for-docker — Docker/container security
[status] security-review-skill-for-terraform — Terraform/IaC security
[status] sca-ai-denoise — SCA vulnerability denoising
[status] supply-chain-audit — supply chain poisoning detection
[status] skills-audit — third-party Skill audit
[status] web-vuln-analyzer — web vulnerability analysis
[status] android-vuln-analyzer — Android vulnerability analysis
[status] security-review-skill-for-* — project-specific security audit
🎨 Development Aids:
[status] frontend-design — frontend UI generation
[status] canvas-design — visual design
📝 Project Skills (.claude/skills/):
[list project-level Skills]
Status labels: ✅ Installed | ❌ Not installed | ⚠️ Needs configuration
High-Frequency Scenarios
| Scenario | Action |
|---|---|
| Initialize a new project | Say "harness" |
| Start a new feature | Say "/plan" → use superpowers brainstorming |
| Debug a bug | superpowers auto-injects systematic-debugging |
| Security audit | "Generate a security audit skill for this project" → security-review-skill-creator |
| SCA denoising | "Analyze these SCA vulnerabilities" → sca-ai-denoise |
| Supply chain audit | "Check dependency security" → supply-chain-audit |
| Docker security review | "Review Docker configuration" → security-review-skill-for-docker |
| Pre-commit check | Say "quality gate" → harness-quality-gate |
| Check project health | Say "harness audit" |
| Clean up temp files | Say "harness cleanup" → interactive archive to archive/ |
| Resume after /compact | Say "harness resume" → lightweight recovery (~3k tokens) |
| Hand off to new agent | Say "harness handoff" → deep handoff (~8k tokens) |
| Recommend a Skill | Say "harness guide" or "which skill should I use" |
| Capture experience | Say "/claudeception" |
| Create a new Skill | "Help me create a skill for XX" → skill-creator |
Re-initialization
If you need to reconfigure (e.g., add Skills, update hooks):
- Re-run
harness— idempotent execution, only supplements missing parts, never overwrites existing content