Harness Model Router
Use when harness-orchestrator is about to dispatch parallel subagents, to classify each task by complexity and annotate it with the right model tier (haiku/sonnet/opus) so cheaper models handle trivial work automatically.
Harness Analyzer
Use at the start of harness work in a new project, or when project grounding is missing or stale, to scan a codebase and GENERATE project-specific grounding skills, Mermaid diagrams, and a relevance index under .claude/skills/, then wire CLAUDE.md so future feature requests automatically follow the harness flow. Run this once before building features.
Harness Verifier
Use after generating or changing code in a harness task to verify it against quality gates — run the project's real lint, type-check, format, and test commands, and check the code matches the project's naming/structure/import conventions. Reports pass/fail and drives fixes until green before the work is considered done.
Harness Context Loader
Use before generating or modifying code in a harness-grounded project to assemble a compact per-loop context packet — pull the loop state (.claude/harness/state.md) plus only the project-* skills and diagrams the current task needs (via project-index.md) into a focused brief with acceptance criteria. Keeps each loop's context precise.
Harness Orchestrator
Use to drive a multi-step build from a requirement — decompose it into tasks and run each through the harness loop (ground → plan → build → verify → test). Coordinates the other harness skills and tracks progress. Invoke when the user asks to implement a feature "with the harness" or wants grounded, verified, end-to-end generation.
Verify
A Claude skill from chetan25/harness-claude-skills.