Claude Cursor Skill

audit-code

Run a two-pass, multidisciplinary code audit led by a tie-breaker lead, combining security, performance, UX, DX, and edge-case analysis into one prioritized report with concrete fixes. Use when the user asks to audit code, perform a deep review, stress-test a codebase, or produce

LLM Mart · 0 points · 4 views 0 listing impressions 0 install-command copies
Download aaaaqwq-agi-super-team-skills_agent-skills-audit-cdb04e8.zip · 9 KB

Install

skills CLI npx skills add https://github.com/aAAaqwq/AGI-Super-Team/tree/main/skills/agent-skills-audit
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install aaaaqwq-agi-super-team@llmmart
Git git clone https://github.com/aAAaqwq/AGI-Super-Team.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole aaaaqwq/agi-super-team collection as a plugin from our marketplace. Git is the plain clone.

README

audit-code skill

Single-source SKILL.md repository for multi-agent distribution.

What this skill is

audit-code is a structured, two-pass multidisciplinary code-audit skill. It guides an expert-panel audit (security, performance, UX, DX, edge cases) with tie-breaker reconciliation and a prioritized final report.

What this skill does

  • Produces findings-first audits with severity, confidence, evidence, and concrete fixes.
  • Uses an invariant coverage matrix to catch cross-route parity issues.
  • Emphasizes operationally realistic verification plans.
  • Includes Bun + SQLite heuristics for common hidden failure modes.

Goal

Keep one canonical skill repository and install/sync it into:

  • Codex: ~/.codex/skills/<skill-name>
  • Claude Code: ~/.claude/skills/<skill-name>
  • Cursor: ~/.cursor/skills/<skill-name>

Why this layout

  • Claude Agent Skills are filesystem-based directories centered on SKILL.md with optional resources.
  • Cursor Agent Skills are also SKILL.md-based and dynamically loaded.
  • Open skills tooling (npx skills) supports multiple agents and both copy and symlink install methods.

Given current ecosystem behavior, copy is the safest default for broad compatibility (especially for global/home installs), while symlink remains optional for local developer workflows.

Sync script

Use the included script to fan out from this canonical repo:

scripts/sync-to-agents.sh

Options:

# default: copy to codex, claude, cursor
scripts/sync-to-agents.sh

# symlink mode (optional)
scripts/sync-to-agents.sh --method symlink

# subset of agents
scripts/sync-to-agents.sh --agents codex,claude

Optional: install via npx skills

If this repo is published remotely, you can install/update with the open installer:

npx skills add <owner>/<repo> -a codex -a claude-code -a cursor --method copy

(Use --method symlink only when you explicitly want linked working copies.)

Recommended workflow

  1. Edit only this canonical repo.
  2. Run scripts/sync-to-agents.sh after changes.
  3. Keep copy as default for portability.
  4. Use symlink only for local development convenience.

References

Skill manifest

Audit Code

Overview

Run an expert-panel audit with strict sequencing and one unified output document. Produce findings first, sorted by severity, with file references, exploit/perf/flow impact, and actionable fixes.

Load references/audit-framework.md before starting the analysis.

Required Inputs

Collect or infer the following:

  • Audit scope: paths, modules, PR diff, or whole repository.
  • Product context: PRD/spec/user stories, trust boundaries, and critical business flows.
  • Runtime context: deployment model, queue/cron/background jobs, traffic profile, data sensitivity, and abuse assumptions.
  • Constraints: timeline, acceptable risk, and preferred remediation style.

If product context is missing, state assumptions explicitly and continue.

Team Roles

Use exactly these roles:

  • Security expert
  • Performance expert
  • UX expert
  • DX expert
  • Edge case master
  • Tie-breaker team lead

The tie-breaker lead resolves conflicts, prioritizes issues, and produces the final single report.

Workflow

Follow this sequence every time:

  1. Build Context Read code + product flows. Identify assets, entry points, high-risk operations, privileged actions, external dependencies, and "failure hurts" journeys.

  2. Build Invariant Coverage Matrix Before specialist pass 1, map critical invariants to every mutating path (HTTP routes, webhooks, async jobs, scripts):

  • Data-link invariants: multi-table relationships that must remain consistent.
  • Auth lifecycle invariants: disable/revoke semantics for sessions/tokens/API keys.
  • Input/transport invariants: validation, content-type policy, body-size/parse behavior.
  • Shape invariants: trees/graphs must reject cycles where applicable. Treat missing parity across equivalent paths as a finding candidate.
  1. Pass 1 Specialist Reviews Run role-specific analysis in this order:
  • Security
  • Performance
  • UX
  • DX
  • Edge case master Capture findings using the schema in references/audit-framework.md.
  1. Tie-Breaker Reconciliation Resolve disagreements:
  • Decide whether contested items are true issues.
  • Set severity and confidence.
  • Remove duplicates and merge overlapping findings.
  1. Cross-Review Pass 2 After edge-case findings, rerun specialists:
  • Security/Performance/UX/DX reassess prior findings and new edge-triggered scenarios.
  • Edge case master performs a final pass on residual risk after proposed mitigations.
  1. Final Report Publish one document from the tie-breaker lead with:
  • Findings first (ordered by severity, then blast radius, then exploitability).
  • Open questions/assumptions.
  • Remediation plan with priority, owner type, and verification tests.
  • Short executive summary at the end.

Quality Bar

Enforce these requirements:

  • Use concrete evidence with file references and line numbers where available.
  • Include reproduction steps for security/performance/edge findings when feasible.
  • Prefer actionable fixes over abstract advice.
  • Separate confirmed defects from speculative risks.
  • Mark confidence for each finding.
  • Run a cross-route consistency sweep: equivalent endpoints/jobs must enforce equivalent invariants.
  • For each High/Critical finding, include at least one focused regression test/check.

Safety and Policy Guardrails

Apply these guardrails while auditing:

  • Do not provide operational abuse instructions or exploit weaponization details.
  • Evaluate manipulative UX patterns as legal/trust/reputation risk, not as recommended growth tactics.
  • Prioritize user safety, system integrity, and maintainable engineering outcomes.

Output Format

Follow this response structure:

  1. Findings List only validated issues. Use the finding schema in references/audit-framework.md.

  2. Open Questions / Assumptions State missing context that could change priority or validity.

  3. Change Summary Summarize high-impact remediation themes in a few lines.

  4. Suggested Verification List focused tests/checks to confirm each major fix.

Runtime Heuristics

When the target stack is Bun + SQLite, apply the runtime-specific checklist in references/audit-framework.md (Runtime-Specific Heuristics (Bun + SQLite)) before finalizing findings.

Files (agi-super-team)
  • .clawhub
    • origin.json 150 B
      Loading…
  • agents
    • openai.yaml 296 B
      Loading…
  • references
    • audit-framework.md 7.2 KB
      Loading…
  • scripts
    • sync-to-agents.sh 2.8 KB
      Loading…
  • README.md 2.5 KB
    # audit-code skill
    
    Single-source `SKILL.md` repository for multi-agent distribution.
    
    ## What this skill is
    
    `audit-code` is a structured, two-pass multidisciplinary code-audit skill. It guides an expert-panel audit (security, performance, UX, DX, edge cases) with tie-breaker reconciliation and a prioritized final report.
    
    ## What this skill does
    
    - Produces findings-first audits with severity, confidence, evidence, and concrete fixes.
    - Uses an invariant coverage matrix to catch cross-route parity issues.
    - Emphasizes operationally realistic verification plans.
    - Includes Bun + SQLite heuristics for common hidden failure modes.
    
    ## Goal
    
    Keep one canonical skill repository and install/sync it into:
    - Codex: `~/.codex/skills/<skill-name>`
    - Claude Code: `~/.claude/skills/<skill-name>`
    - Cursor: `~/.cursor/skills/<skill-name>`
    
    ## Why this layout
    
    - Claude Agent Skills are filesystem-based directories centered on `SKILL.md` with optional resources.
    - Cursor Agent Skills are also `SKILL.md`-based and dynamically loaded.
    - Open skills tooling (`npx skills`) supports multiple agents and both `copy` and `symlink` install methods.
    
    Given current ecosystem behavior, **copy is the safest default** for broad compatibility (especially for global/home installs), while symlink remains optional for local developer workflows.
    
    ## Sync script
    
    Use the included script to fan out from this canonical repo:
    
    ```bash
    scripts/sync-to-agents.sh
    ```
    
    Options:
    
    ```bash
    # default: copy to codex, claude, cursor
    scripts/sync-to-agents.sh
    
    # symlink mode (optional)
    scripts/sync-to-agents.sh --method symlink
    
    # subset of agents
    scripts/sync-to-agents.sh --agents codex,claude
    ```
    
    ## Optional: install via `npx skills`
    
    If this repo is published remotely, you can install/update with the open installer:
    
    ```bash
    npx skills add <owner>/<repo> -a codex -a claude-code -a cursor --method copy
    ```
    
    (Use `--method symlink` only when you explicitly want linked working copies.)
    
    ## Recommended workflow
    
    1. Edit only this canonical repo.
    2. Run `scripts/sync-to-agents.sh` after changes.
    3. Keep `copy` as default for portability.
    4. Use `symlink` only for local development convenience.
    
    ## References
    
    - Claude docs: https://platform.claude.com/docs/en/agents-and-tools/agent-skills/overview
    - Cursor docs entry point: https://cursor.com/docs/context/skills
    - Cursor official usage details: https://cursor.com/blog/agent-best-practices and https://cursor.com/changelog/2-4
    - Open skills installer: https://github.com/vercel-labs/skills
    
  • SKILL.md 4.5 KB
    ---
    name: audit-code
    description: Run a two-pass, multidisciplinary code audit led by a tie-breaker lead, combining security, performance, UX, DX, and edge-case analysis into one prioritized report with concrete fixes. Use when the user asks to audit code, perform a deep review, stress-test a codebase, or produce a risk-ranked remediation plan across backend, frontend, APIs, infra scripts, and product flows.
    ---
    
    # Audit Code
    
    ## Overview
    
    Run an expert-panel audit with strict sequencing and one unified output document.
    Produce findings first, sorted by severity, with file references, exploit/perf/flow impact, and actionable fixes.
    
    Load `references/audit-framework.md` before starting the analysis.
    
    ## Required Inputs
    
    Collect or infer the following:
    - Audit scope: paths, modules, PR diff, or whole repository.
    - Product context: PRD/spec/user stories, trust boundaries, and critical business flows.
    - Runtime context: deployment model, queue/cron/background jobs, traffic profile, data sensitivity, and abuse assumptions.
    - Constraints: timeline, acceptable risk, and preferred remediation style.
    
    If product context is missing, state assumptions explicitly and continue.
    
    ## Team Roles
    
    Use exactly these roles:
    - Security expert
    - Performance expert
    - UX expert
    - DX expert
    - Edge case master
    - Tie-breaker team lead
    
    The tie-breaker lead resolves conflicts, prioritizes issues, and produces the final single report.
    
    ## Workflow
    
    Follow this sequence every time:
    
    1. Build Context
    Read code + product flows. Identify assets, entry points, high-risk operations, privileged actions, external dependencies, and "failure hurts" journeys.
    
    2. Build Invariant Coverage Matrix
    Before specialist pass 1, map critical invariants to every mutating path (HTTP routes, webhooks, async jobs, scripts):
    - Data-link invariants: multi-table relationships that must remain consistent.
    - Auth lifecycle invariants: disable/revoke semantics for sessions/tokens/API keys.
    - Input/transport invariants: validation, content-type policy, body-size/parse behavior.
    - Shape invariants: trees/graphs must reject cycles where applicable.
    Treat missing parity across equivalent paths as a finding candidate.
    
    3. Pass 1 Specialist Reviews
    Run role-specific analysis in this order:
    - Security
    - Performance
    - UX
    - DX
    - Edge case master
    Capture findings using the schema in `references/audit-framework.md`.
    
    4. Tie-Breaker Reconciliation
    Resolve disagreements:
    - Decide whether contested items are true issues.
    - Set severity and confidence.
    - Remove duplicates and merge overlapping findings.
    
    5. Cross-Review Pass 2
    After edge-case findings, rerun specialists:
    - Security/Performance/UX/DX reassess prior findings and new edge-triggered scenarios.
    - Edge case master performs a final pass on residual risk after proposed mitigations.
    
    6. Final Report
    Publish one document from the tie-breaker lead with:
    - Findings first (ordered by severity, then blast radius, then exploitability).
    - Open questions/assumptions.
    - Remediation plan with priority, owner type, and verification tests.
    - Short executive summary at the end.
    
    ## Quality Bar
    
    Enforce these requirements:
    - Use concrete evidence with file references and line numbers where available.
    - Include reproduction steps for security/performance/edge findings when feasible.
    - Prefer actionable fixes over abstract advice.
    - Separate confirmed defects from speculative risks.
    - Mark confidence for each finding.
    - Run a cross-route consistency sweep: equivalent endpoints/jobs must enforce equivalent invariants.
    - For each High/Critical finding, include at least one focused regression test/check.
    
    ## Safety and Policy Guardrails
    
    Apply these guardrails while auditing:
    - Do not provide operational abuse instructions or exploit weaponization details.
    - Evaluate manipulative UX patterns as legal/trust/reputation risk, not as recommended growth tactics.
    - Prioritize user safety, system integrity, and maintainable engineering outcomes.
    
    ## Output Format
    
    Follow this response structure:
    
    1. Findings
    List only validated issues. Use the finding schema in `references/audit-framework.md`.
    
    2. Open Questions / Assumptions
    State missing context that could change priority or validity.
    
    3. Change Summary
    Summarize high-impact remediation themes in a few lines.
    
    4. Suggested Verification
    List focused tests/checks to confirm each major fix.
    
    ## Runtime Heuristics
    
    When the target stack is Bun + SQLite, apply the runtime-specific checklist in `references/audit-framework.md` (`Runtime-Specific Heuristics (Bun + SQLite)`) before finalizing findings.
    
  • _meta.json 137 B
    Loading…

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.