Auto Harness Skill

auto-harness is a Codex skill for diagnosing and strengthening a repository's harness layer.

It focuses on the project-specific control system around product code:

• AGENTS.md rules and repo operating guidance
• architecture boundaries and ownership seams
• lint, type, and import-layer guardrails
• API and generated-client contracts
• CLI/backend/frontend drift prevention
• test scaffolding, regression strategy, per-language verification calibration, and structured knowledge/doc alignment
• structured logging for expensive external integrations
• technical-debt and migration tracking

Repository layout

skills/
  auto-harness/
    SKILL.md
    agents/openai.yaml
    references/

Install with Codex

From GitHub:

python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
  --repo PacificStudio/auto-harness-skill \
  --path skills/auto-harness

Or via URL:

python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
  --url https://github.com/PacificStudio/auto-harness-skill/tree/main/skills/auto-harness

Then restart Codex.

What the skill provides

• a harness diagnosis workflow
• a project audit checklist
• concrete guardrail patterns
• a staged fix-plan template
• a reusable set of drift-audit passes for duplicated authority, incomplete projections, structured vs freeform drift, cross-surface contracts, runtime/doc/tooling mismatches, language verification matrices, knowledge topology, and bidirectional doc/code alignment
• a non-guess verification lens for coverage realism, CI wiring, and environment completeness across Go, TS/JS, Python, Java, shell, and mixed-runtime repos
• guardrail patterns for semantic knowledge trees, structured doc/code binding manifests, pre-push doc review gates, and file-budget checks for docs and frontend-heavy files

The actual skill content lives in skills/auto-harness/.