Plugin Claude Code
Autonomous security auditor. Scans a GitHub repo for vulnerabilities, triages false positives, writes a PoC, fixes each confirmed bug in its own PR, independently reviews the fix, and merges when the review is clean.
Agent
Build a minimal proof of concept for a confirmed security finding. Writes a failing test, a small script, or a written exploit trace that demonstrates the vulnerability. Does NOT modify source code. Invoke this when a finding is in the confirmed state.
Agent
Implement the minimal fix for a confirmed security finding with a working PoC. Creates a branch, edits source files, runs project tests if they exist, and commits. Does NOT open the PR. Invoke when a finding is in pocwritten state.
Agent
Independently review the fix PR for a security finding. Gets ONLY the raw finding description and the diff — no access to the triage or fixer's reasoning. Decides approve or requestchanges. Invoke when a finding is in propened state.
Agent
Independently evaluate a single security finding to decide whether it is a real, exploitable vulnerability or a false positive. Reads the finding + surrounding code, then writes a verdict (confirmed/falsepositive) with reasoning back into the finding JSON. Invoke this when a finding is in the discovered state.
Hook
Runs when a session starts on resume and startup, executing session-start.sh. From wrxck/auto-audit.
Skill Claude CodeCodex
Scan the active auto-audit workspace for security vulnerabilities and add any new findings to the queue. Run as part of /auto-audit:start or periodically via /auto-audit:tick rescans. Uses LLM-based code review plus available CLI scanners (npm audit, pip-audit, gitleaks-like regexes).
Skill Claude CodeCodex
Install the auto-audit badges in the target repo. Publishes the live status JSON to the autoaudit/status branch so the dynamic shields.io badge renders, and optionally opens a PR adding the static 'audited by auto-audit' badge to the README. Use after an audit when the user wants visibility in their repo's README.
Skill Claude CodeCodex
Record operator feedback against the active repo's audit so future triager and fixer runs incorporate it. Use when the user says 'don't write that pattern again', 'I reverted that PR because X', 'the triager got this one wrong', or 'remember this for next time'. The reviewer is independent and does NOT read this log …
Skill Claude CodeCodex
Generate a self-contained HTML audit report for the active repo (or a named one). The report includes summary stats, per-finding triage / PoC / fix / review detail, and the full activity log. Use when the user says 'generate a report', 'export findings', 'build the audit summary', 'write up the audit'.
Skill Claude CodeCodex
Resume a stopped or previously-interrupted autonomous audit without re-initialising the workspace. Re-establishes the active repo pointer and kicks off the loop again. Use when the user says 'resume auto-audit', 'continue the audit', or after a session restart.
Skill Claude CodeCodex
Autonomously audit a GitHub repo. Clones the repo, runs a full audit scan (security by default), then kicks off a continuous loop that triages findings, writes proofs of concept, fixes issues in PRs, independently reviews, and merges. Use when the user says things like 'audit my repo', 'scan repo X for security…
Skill Claude CodeCodex
Show the current state of the autonomous audit: active repo, findings breakdown by status, recent activity, next pending finding. Use when the user asks 'what's auto-audit doing?', 'audit status', 'audit progress'.
Skill Claude CodeCodex
Halt the autonomous audit loop. Drops the active-repo pointer so ticks become no-ops, and advises the user to press Esc to cancel the running /loop. Use when the user says 'stop auto-audit', 'halt the audit', 'cancel'.
Skill Claude CodeCodex
Advance the autonomous audit by one finding. Picks the next pending finding, moves it through the next lifecycle stage (triage, PoC, fix, PR, review, merge), and stops. Intended to be called repeatedly by /loop — do not use this directly; use /auto-audit:start instead.
Plugin Claude Code
Learns from your corrections and writes hooks so Claude never repeats the same mistake.
Skill Claude CodeCodex
Detects user corrections and frustrations, then writes enforceable hooks so the same mistake never happens again. Triggers on phrases like "don't do that", "stop doing", "I told you not to", "why did you", "that's wrong", or any expression of displeasure with Claude's behavior.
Plugin Claude Code
Create detailed Mermaid diagrams for architecture, flows, ERDs, and more — and improve READMEs with visual documentation.
Agent
Analyses project structure, dependencies, and flows to suggest appropriate diagrams. Use when needing to understand a codebase before generating diagrams.
Command
Generate a Mermaid diagram from a description or by analysing the current codebase.
Skill Claude CodeCodex
Create detailed, accurate Mermaid diagrams for software architecture, data flows, ERDs, sequences, and more. Improve READMEs with visual documentation.
Plugin Claude Code
Curated Claude Code plugins by Matt Hesketh.
Plugin Claude Code
Scaffold hardened MCP servers wrapping CLI tools.
Plugin Claude Code
Make da agent talk Minionese - conversational prose only, code and commands stay exact.