Claude Code only. Use when the user explicitly invokes /agents-bridge:codex-orchestrate (or /codex-orchestrate) in Claude Code or asks Claude to orchestrate a large multi-slice plan by delegating implementation to the Codex CLI through agents-bridge. In native Codex, use $slice-runner instead.
Critical second opinion from Codex (OpenAI) on a proposal Claude just made — confirms what is sound, challenges what is genuinely weak, and suggests a better path with clear reasoning when one exists. Constructive, not contrarian. Use to cross-check a Claude design, refactor, API, or fix with a non-Claude model before…
Generate or edit images with the Codex CLI's built-in imagegen tool. Use when the user asks to create, generate or draw an image, icon, logo, favicon, illustration or mockup, or to edit an existing image file — restyle, retouch, upscale, add or remove elements.
Start a session in pair-planning mode where Claude and Codex (read-only) each draft an independent implementation plan for a task, then cross-review through an open-point ledger until consensus, escalating remaining disagreements to the user. Produces one agreed plan; does not write code. Use at the start of a task to…
Audit an Effect codebase for silent error swallowing -- catchAll patterns that collapse errors into sentinel values (null, [], false, void) in the success channel. Runs a scanner script, spawns a read-only auditor, and produces a classified report. NOT for: non-Effect projects, React/frontend code, plain TypeScript…
Audit a codebase for handcrafted code that duplicates functionality already available in the project's dependencies. Reads package.json, launches parallel exploration agents, verifies replacement feasibility, and produces a structured refactor plan. Audit only -- does not execute changes.
Rigorous planning workflow with research validation, task breakdown, dependency mapping, and post-implementation verification. Use when: (1) planning complex features touching 5+ files, (2) architectural changes or major refactors, (3) unfamiliar tech requiring doc research, (4) user says "thorough plan", "rigorous…
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
Audit and clean code comments with four actions: REFACTOR (comment compensates for unclear code), KEEP (why / external constraint / trap), REMOVE (paraphrase, decorative banners), FIX-VERIFY (factual claim checked against the code). Protects why and constraint comments; hunts comments that lie — stale cited paths…
Verify no behavioral regressions after implementation. Spawns parallel subagents for semantic diff review + test execution. Use when: user says "verify no regressions", "check for regressions", "make sure nothing broke".
Thorough planning for complex features via delegated subagents. Turn discussions into executable, self-contained implementation plans. USE for multi-file features, architectural changes, unfamiliar tech requiring research. SKIP for quick fixes or single-file changes.
Audit and improve CLAUDE.md, AGENTS.md and .claude/rules/ for instruction budget, anti-patterns and stale references, then apply approved REMOVE/MOVE/REWRITE/ADD fixes.
Clean stale git branches and worktrees interactively. Audits local/remote branches, proves which ones the base already contains (including after a squash or rebase), frees finished worktrees, and confirms before deleting. Use when: "git sweep", "sweep branches", "branch cleanup", "prune stale branches", "clean up…