菜单
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.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Multi-pass adversarial performance audit for entire repositories. Combines structured profiling (Agent A) with adversarial stress-testing critique (Agent B) through iterative passes. Optimize repo/backend performance, hot-paths, build size, and server throughput. Do NOT use for frontend page load metrics or Lighthouse audits (use web-perf). NOT for frontend Core Web Vitals (use web-perf).
Multi-pass adversarial planning and review skill that improves agent-generated plans through structured critique stages. Combines an initial planning agent (structure, logic, task sequencing) with an adversarial review agent (performance, security, maintainability) and a final Copilot extension validation pass. Use when creating implementation plans, designing architecture, planning roadmaps or milestones, or when the user says "plan this", "review my plan", "adversarial review", or "multi-pass plan".
Multi-pass adversarial security audit for entire repositories. Combines structured threat modeling (Agent A) with adversarial attack surface analysis (Agent B) through iterative passes. Merges the security-audit workflow's 10-category checklist with the adversarial-planner's structured critique methodology. Use when running security audits, threat modeling, or when the user says "security audit", "adversarial security", "threat model this repo", "red team this repo", or "find vulnerabilities". NOT for fixing or remediating known security issues (use autonomous-dev instead). NOT for supply chain dependency scanning. NOT for general workflow audits or code quality. If user says 'fix security issues', ask whether they want to audit first or fix known issues.
Multi-pass adversarial quality audit for flat markdown workflow files. Combines structured evaluation (Agent A) with adversarial stress-testing (Agent B) to assess sequential clarity, prerequisite validation, HITL (Human-in-the-loop) safety gates, loop prevention, and formatting. Use when auditing workflow scripts (e.g., slash-commands like /bump-deploy), or when the user asks to review or improve operational playbooks.
Core rules for code quality and specifications of Model Context Protocol (MCP) servers. Use when reviewing MCP code quality, enforcing specification rules, or checking schemas/error responses. Must see the explicit keyword "MCP" or "Model Context Protocol". NOT for general REST APIs or Cloudflare Workers. Do NOT trigger for generic 'build a server' requests unless the type is explicitly specified as an MCP server. (Invoke explicitly with `/mcp-builder` — does not auto-trigger).
Build AI agents on Cloudflare Workers using the Agents SDK or build Model Context Protocol (MCP) servers. Load when creating stateful agents, durable workflows, MCP servers, checking MCP schema/error responses, or reviewing MCP code quality. Covers Agent class, state management, callable RPC, Workflows integration, and MCP implementation rules. Do NOT trigger for generic 'build a server' requests unless the platform is explicitly specified as Cloudflare Agents or an MCP server.
| 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. |
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.
Load this skill when any of these apply:
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
The 8 quality categories evaluated per skill:
Additionally, 4 directory-level categories assess the collection:
For the full checklist, read references/audit-categories.md.
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.
Every phase creates a journal entry for future retrieval. For templates and tag conventions, read references/feedback-loop.md.
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.| 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 |
skill-builder/checklist.md item by item| 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 |