claude-code-zero

Set up worktree-plus for a repo — manage git config (baseBranch, branchPrefix, dirBase, guessRemote) and build the .worktreeinclude / .worktreelink files by scanning which gitignored files actually exist. Use when the user asks to set up or configure worktree-plus, view or change worktree settings, reset the plugin config, or decide which gitignored files should be copied or symlinked into new worktrees.

Delegate an implementation task to Codex, then Claude reviews the result. Use when asked "codex rescue", "delegate to codex", "have codex do it", or wants Codex to implement or fix something.

Check Codex CLI, auth, Official plugin status, and configure defaults. Use when asked "codex setup", "configure codex", "install codex", or when another codex-advisor skill reports setup issues.

Diagnose Claude Code environment health — context budget, description obesity, trigger collisions, hooks, MCP, plugins, CLAUDE.md, memory, and skill-security scan. Use when asked to audit the environment, check context budget, review plugins, or scan installed skills for risky patterns.

Visualize git diffs as interactive HTML reports with architecture diagrams, change analysis, and side-by-side split-diff of the actual changed code. Use when asked to visualize, review, or summarize a diff, branch, commit, or PR — including seeing the real changed lines, not just a summary. Accepts branch names, commit hashes, HEAD, PR numbers, or commit ranges.

Convert any markdown document into a diagram-enriched visual HTML report. Use when asked to visualize, illustrate, or create a visual report from a document with diagrams. Single md file input. Also trigger when: "make this document visual", "add diagrams to this doc", "turn this markdown into a report", "visualize this README/ADR/spec", or any request to render a document with embedded diagrams.

Analyze agent extensions and generate self-contained HTML wiki reports with security audit and architecture diagrams. Use when asked to analyze, audit, or document a plugin. Triggers on GitHub plugin URLs or local plugin paths.

Verify factual accuracy of a document against the codebase and git history. Use when asked to verify, fact-check, or audit claims in a report or document. Accepts a file path or auto-detects the most recent HTML report.

Codebase-orientation session with the rubber duck — generate or refresh .claude/orientation.md with interactive exercises. Use when new to a codebase, returning after a break, or says "duck orient", "I'm new to this repo".

Senior-engineer coaching session: explains a topic in plain language with one concrete analogy then walks example → exercise → critique; or dissects code just generated in this session section-by-section (what it does / why this way over the obvious alternative / where it breaks first in prod); or surfaces an unresolved gap from a past duck session and teaches it. Manual-invocation only — call with /coach <topic>, /coach <file>, or /coach alone, e.g. "/coach embeddings" or "/coach src/foo.ts".

Manage and refine vision-powers reports: list, open, delete, search, and refine sections. Use when asked to list, open, delete, search, or update generated HTML reports.

Code-verification session with the rubber duck — user explains code just written, finds edge cases, fixes planted bugs. Use after implementing a feature, or when they say "duck verify", "double check this". Not for plan review (/duck-prebuild) or PR review (/duck-review).

Pre-build session with the rubber duck — runs BEFORE implementation, covering both a pre-coding design sketch and plan/spec review. Use when the user states implementation intent ("build this for me", "build me X") with no plan or code yet, when a plan/spec/RFC was just produced, or says "duck design", "duck plan", "sketch first", "review this plan". Not for code already written — use /duck-verify or /duck-review.

PR/change-review session with the rubber duck — user justifies every change and predicts consequences. Use before commit/push/PR, or when they say "duck review", "check before review". Not for code-level explanation (/duck-verify) or plan review (/duck-prebuild).

Rubber duck tutor — auto-detects which review mode to run. Use when the user says "duck", "tutor", "quiz me", "do I understand this", or mentions rubber-stamping or learning while coding. For a specific phase use /duck-prebuild, /duck-verify, /duck-review, or /duck-orient.

Hand off the current Claude Code session to Codex as a resumable thread with full turn history, then continue outside Claude entirely. Use when asked "codex transfer", "move this session to codex", "continue this in codex", "hand this off to codex", or wants to migrate the conversation itself (not delegate one task and come back — for that use /codex-rescue).

Run Codex adversarial review — actively tries to break confidence in the change. Use when asked "adversarial review", "red-team this change", or wants thorough security/correctness challenge.

Run Codex code review with Claude's independent double-check. Use when asked "codex review", "review my code with codex", or wants Codex to review code changes. For adversarial review use /codex-adversarial.

Create new skills, modify and improve existing skills, and measure skill performance for Claude Code. Use when users want to create a skill from scratch, edit or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy. Also trigger on "make a skill", "skill for X", "improve my skill", "turn this into a skill", or capturing a workflow as a reusable skill.

Deep-dive research using Codex with Claude's cross-model synthesis. Use when asked "codex research", "deep dive with codex", "investigate this topic". Not for code review or plan verification.

Verify a plan or document using Codex as independent reviewer with PASS/FAIL verdict. Use when asked "codex verify", "verify this plan", "review this doc for issues".

Cancel an active background Codex job. Use when asked "codex cancel", "stop codex", "abort the codex job", or wants to stop a running Codex task.

Show the final stored result of a completed Codex job. Use when asked "codex result", "show the codex output", "show me the job result", or wants output from a finished job.

List active and recent Codex jobs plus stored review files. Use when asked "codex status", "what codex jobs are running", or wants to see Codex job state.

Distill session into a resumption-ready handoff document

Autonomously optimize any Claude Code skill by running it repeatedly, scoring against binary evals, mutating the prompt, and keeping improvements. Use when: optimize/improve/benchmark/eval a skill, autoresearch, auto-optimize. Not for creating skills from scratch (use skill-creator-pro).

Stop the mo markdown viewer server for the current project. Use when the user says claw-mo-down, wants to stop mo, shut down doc viewer, or kill mo server.

Configure mo markdown viewer for the current project. Use when the user wants to set up doc watching, configure mo patterns, initialize claw-mo for a project, or set up markdown viewer groups.

Start or reopen the mo markdown viewer for the current project. Use when mo is already configured and the user wants to view docs, start mo, open documentation viewer, or preview markdown.

Pull latest changes from origin for all git repos under a directory. Use when asked to update, sync, or pull reference projects.

Discover, choose, and rewrite backend-specific VibeProxy cc-* aliases for Claude Code. Use when setting up, resetting, or auditing VibeProxy aliases for Codex, Copilot, Antigravity, Gemini, Qwen, or Z.AI GLM.

Manage mo markdown viewer — check status, add/remove patterns and groups, refresh, stop servers, reset. Use when checking mo status, modifying watch patterns, or troubleshooting.

Control cmux terminal topology and I/O — send commands to panes, read output, split layouts, monitor logs, orchestrate multi-pane workflows. Requires cmux environment.

Manages NotebookLM notebooks — query, add, list, search, enable/disable, remove. Use when a notebooklm.google.com URL appears or user mentions NotebookLM. Handles URLs through its own Chrome agent — do not navigate directly. Requires: claude-in-chrome MCP.

Open a specific markdown file, directory, or piped content in mo. Use when the user wants to quickly view one markdown file, add a file to mo, pipe markdown to the viewer, or preview a specific document.

Browser automation with cmux embedded webviews (not Chrome/Chromium). Use when $CMUX_WORKSPACE_ID is set and you need to open sites, interact with pages, wait for state changes, or extract data. Snapshot/ref workflow for reliable element targeting.

Extract hardcoded secrets from CLAUDE.md, .mcp.json, and project config into a gitignored env file, then wire a SessionStart hook to load them automatically. Use when user says "separate secrets", "extract API keys", "secret setup", "env var setup", "hardcoded credentials", ".env setup for Claude", "load secrets via hook", "protect credentials", or "clean up mcp secrets".

Open markdown files in a cmux formatted viewer panel with live reload. Use when you need to display plans, documentation, or notes alongside the terminal with rich rendering (headings, code blocks, tables, lists). Requires cmux environment ($CMUX_WORKSPACE_ID).

Run E2E browser tests from natural language JSON test files using agent-browser. Use when asked to run e2e tests, browser tests, UI tests, end-to-end tests, or test a web application.

Extract URLs from an XML sitemap with optional regex filtering