adversarial-skill-audit

name	adversarial-skill-audit
description	Multi-pass adversarial quality audit for agent skill directories. Combines structured evaluation with adversarial stress-testing to assess skill completeness, instruction clarity, trigger accuracy, and security. Use when auditing a skills directory.

Adversarial Skill Audit

A multi-pass quality auditing system for agent skill directories. Evaluates each skill against the skill-builder quality standards through structured profiling and adversarial stress-testing — finding gaps in trigger coverage, instruction clarity, progressive disclosure, security, and cross-skill coherence.

When to Load

Load this skill when any of these apply:

• Auditing an entire skills directory for quality and consistency
• Reviewing a batch of skills before publishing or distributing
• The user asks for skill quality review, audit, or improvement suggestions
• The user says "audit my skills", "skill quality check", "review these skills", "check my skills", "how good are my skills", or "validate these skills"
• Preparing a skills directory for npm packaging or distribution
• You want to identify redundant, incomplete, or poorly triggered skills

Adversarial Protocol

This skill follows the standard dual-agent adversarial pattern (Agent A: The Evaluator, Agent B: The Adversarial User). For the core pipeline rules, phase definitions, and agent switching protocols, read: ../adversarial-security/references/adversarial-base-protocol.md

For the skill audit-specific protocol with scoring and templates, read: references/multi-pass-skill-protocol.md

Audit Categories

The 8 quality categories evaluated per skill:

1. Frontmatter & Triggering
2. Instruction Clarity
3. Structure & Progressive Disclosure
4. Output Formats & Templates
5. Edge Cases & Error Handling
6. Security & Safety
7. Token Efficiency
8. Maintenance & Versioning

Additionally, 4 directory-level categories assess the collection:

9. Cross-Skill Coherence
10. Trigger Collision Detection
11. Coverage Gap Analysis
12. Ecosystem Consistency

For the full checklist, read references/audit-categories.md.

External Validation (Phase 4)

Phase 4 triggers an independent validation pass using the GitHub CLI (gh copilot). The copilot subcommand is built into modern gh CLI — no separate extension is needed. This provides a fundamentally different model's perspective on skill quality, catching issues that internal review normalizes.

For prompts, read references/copilot-skill-prompts.md.

Prerequisites: gh CLI v2.x+ with gh auth status passing. If gh copilot is not available, skip Phase 4 gracefully and note the skip in the journal entry.

Read references/copilot-usage.md for critical non-interactive execution requirements.

Feedback Loop

Every phase creates a journal entry for future retrieval. For templates and tag conventions, read references/feedback-loop.md.

Scripts

This skill includes automated helper scripts located in the scripts/ directory:

• scripts/check-skills.ps1: Automated Phase 1 metric gathering (token count, trigger detection).
• scripts/run-copilot.ps1: Automated Phase 4 Copilot validation pipeline.

Configuration

Variable	Default	Description
MAX_AUDIT_PASSES	2	Maximum stress-test cycles (phases 2–3 repeat)
AUDIT_DEPTH	standard	Depth: surface, standard, or thorough
COPILOT_VALIDATION	true	Enable/disable Copilot extension validation phase
INCLUDE_REFERENCES	true	Whether to read and evaluate reference files too

Audit Depth Profiles

• Surface: Frontmatter + structure only (Categories 1, 3, 7). Quick scan for obvious issues. Best for large directories (30+ skills).
• Standard: All 8 per-skill categories + 4 directory-level categories. Default for most audits.
• Thorough: Full audit + extended analysis:
  • Read every reference file and evaluate its quality
  • Construct 3 test prompts per skill and evaluate trigger likelihood
  • Analyze description keyword coverage against real user phrasing
  • Compare against skill-builder/checklist.md item by item
  • Check for stale content (outdated API references, deprecated tools)

Synergies

Skill/Workflow	Relationship
skill-builder	Defines the quality standards this skill audits against
adversarial-planner	Parent pattern — plan-level adversarial review
adversarial-security	Sibling — audits security posture; this audits skill quality
adversarial-performance	Sibling — audits performance; this audits skill quality