audit-domain-harness

name	audit-domain-harness
description	Use when reviewing existing domain-harness registry, specs, evals, scaffold plans, and project AI configuration for gaps, conflicts, and risk.
metadata	{"short-description":"Audit domain harness artifacts"}

Audit Domain Harness

Workflow

1. Perform a read-only audit of docs/domain-harness/index.md, referenced specs, evals, scaffold plans, and relevant AI configuration.
2. Compare registry status with actual files and activation state.
3. Check whether development, non-development, and mixed guardrails are complete for the declared work_type.
4. Identify conflicts between skills, subagents, rules, hooks, MCP guidance, and project instructions.
5. Report findings before summaries.

Use scripts/validate_domain_harness.py <project-root> only as a read-only audit aid for docs/domain-harness/**; it does not replace evidence-based findings or user-approved remediation.

Output

• Findings ordered by severity
• Evidence
• Impact
• Recommended action
• Open questions
• Residual risk

Do not

• Do not modify files during an audit.
• Do not treat missing evidence as proof that a harness is safe.
• Do not ignore non-development risks such as source quality, privacy, brand, tone, or approval flow.
• Do not manage lab evaluation reports or sanitized evaluation cases from this plugin skill.