release

Bootstrap a new project with interactive brainstorming and structure generation

Review recent work against plan and project standards

Run code review via Codex CLI with non-Claude models (GPT-5, o3, Ollama)

Audit a private repo before making it public

Append a retrospective section to the current plan file

Address PR review feedback and advance toward mergeable state

Review health of ~/code projects (git status, tests, remote sync)

Pause to review work with fresh eyes for bugs, missed cases, or simplifications

Review session trajectory and update todo file with lessons learned

The final step before merging a PR. Run after `/reflect`.

Explain the current status line with verbose breakdown and ASCII diagram

Run intelligent dependency updates with batching and learning

Run an automatic turn-based vocal loop (ask aloud, listen, respond, keep listening)

Implement Vercel AI SDK 6 agents with streaming, tool orchestration, and reasoning visibility.

Curate and organize Chronicle memory blocks. Reviews session context, updates recent blocks, and maintains memory coherence across sessions.

Deep exploration of Chronicle memory and worktrees to generate meaningful insights. Spawns Explore subagents to examine actual code and cross-reference with memory blocks.

Use this agent when you need to set up, configure, or verify a DevContainer environment for a project. This includes creating the initial .devcontainer configuration files, setting up security measures like firewalls and filesystem isolation, configuring development tools, and verifying the container environment is properly configured.

Check GitHub notifications and provide a prioritized summary of important items.

Audit codebase quality, docs, tests, and deployment before client handoff.

Deep codebase exploration to understand a problem and create a research artifact

Sleep-time compute orchestrator for team-memory skill. Dispatched by SessionEnd hook. Runs extract → consolidate → reflect pipeline sequentially.

Background listener that records speech, transcribes it, and sends transcripts to the main session.

File-based messaging between agents across any harness. Invoke this skill when you see "📬 unread in .agents/inbox", need to send or read agent messages, or set up an inbox. Triggers on: "📬", ".agents/inbox", "agent inbox", "send message to agent", "check inbox", "agent message".

Analyze AI coding assistant usage patterns across Claude Code and Cursor. Use when user asks about their coding usage, tool statistics, productivity patterns, skill popularity, session history, or wants to query their AI coding logs. Triggers include "usage", "how much have I used", "most used tools", "skill popularity", "coding stats", "productivity patterns".

Fix misaligned right borders in ASCII art diagrams

Capture explored work as a backlog item for future implementation. Use when you've explored an enhancement, alternative approach, or feature but decided to defer it. Creates comprehensive plan files in backlog/ directory with enough context for a future session to execute efficiently.

Wide→Narrow design workflow taking UI/UX concepts from exploration to polished design brief. Four phases - Context, Explore, Iterate, Refine. Use for visual design exploration, user story mockups, and creating shareable design briefs.

Capture and curate session memory blocks. Use /chronicle to save current work, /chronicle catchup to restore context, /chronicle curate to organize memory, /chronicle insights for deep analysis, /chronicle summarize for AI summaries, /chronicle pending for open threads, /chronicle search to find sessions, /chronicle publish for digests, /chronicle ui for dashboard with repo-level views and usage stats.

Clone GitHub and git repos into a conventional local location, inspect them, and recommend the most useful next steps. Use when the user asks to clone, download, inspect, evaluate, look over, or assess a repository, especially when they want it placed under the usual owner/repo path in ~/code/github and reviewed before deeper work.

Set up Cloudflare Workers deployment for web applications with GitHub Actions CI/CD. Use when adding Workers deployment to an existing project, setting up auto-deploy on push/PR, configuring Durable Objects for stateful apps, or creating a dual-deployment pattern (keep Replit/Vercel while adding Workers). Supports Hono framework, static assets, and nodejs_compat.

Orchestrate the cmux terminal — named layouts (workshop, ops-deck), sidebar dashboarding (status, progress, logs), browser automation, cross-workspace agent coordination, and theme customization. Use this skill whenever inside a cmux terminal (CMUX_WORKSPACE_ID is set), the user mentions cmux, asks to set up a workspace layout, open a browser pane, update sidebar status/progress, spawn agents across workspaces, change themes, or says 'workshop' or 'ops deck'. Also use when other skills need to update the sidebar or rename the workspace tab.

Manage GitHub Codespaces lifecycle via gh CLI. Use for: "create codespace", "ssh into codespace", "stop codespace", "delete codespace", "codespace secrets", "codespace logs", "list codespaces", "codespace ports", "rebuild codespace", "gh cs", "codespace machine types".

Work with Claude Code configuration at global (~/.claude) or project (.claude/) level. Use when editing settings.json (permissions, hooks, statusline, model), managing MCP servers, creating agents/commands/skills, writing CLAUDE.md, setting up rules files, or configuring a new project. Determines context automatically and provides guidance on global vs project placement to avoid duplication.

Create diagrams and visual artifacts using Excalidraw with real-time canvas preview. Use when the user asks to create, draw, or visualize diagrams, flowcharts, architecture diagrams, system designs, or any visual representation. Supports creating elements programmatically, Mermaid diagram conversion, and visual verification via screenshots.

Fork the current session with context carried over. Use `/fork <branch>` for a new worktree or `/fork --local` for a new session in the current directory.

Create distinctive, production-grade frontend interfaces with high design quality. Use this skill when the user asks to build web components, pages, artifacts, posters, or applications (examples include websites, landing pages, dashboards, React components, HTML/CSS layouts, or when styling/beautifying any web UI). Generates creative, polished code and UI design that avoids generic AI aesthetics.

Create, authenticate, and manage GitHub Apps. Use for: creating apps via manifest flow, generating JWTs, getting installation tokens, managing webhooks and installations, rotating keys. Triggers on: "create github app", "github app", "manage github app", "github app JWT", "github installation token", "github app manifest", "github webhook deliveries", "register github app", "gh app".

Manage Git worktrees for concurrent local development. Creates worktrees at ~/.worktrees/REPO/BRANCH. Wrapper for the `wt` CLI.

Generate images using AI APIs (OpenAI gpt-image-1, Google Imagen 4, Nano Banana Pro, fal.ai Flux). Use when user asks to generate, create, or make an image.

Automate iOS Simulator tasks — capture screenshots, interact with apps, generate screen flow galleries. Use when the user asks to capture app screens, screenshot the simulator, document the app UI, create a screen flow, interact with the simulator, or automate iOS testing workflows.

Build and operate a persistent persona and memory framework for Claude Code. Use when setting up personality.md injection, durable memory blocks in ~/.ai-memory, session start/end memory workflows, recall/remember/consolidate scripts, or packaging a distributable memory skill profile.

Standardize project lifecycle scripts (setup, run, stop, archive) in scripts/ so agents can manage workspaces through a single interface across runtimes. Scripts are portable bash; mise is the recommended orchestrator when available, with adapters for Conductor, Claude Code hooks, devcontainers, Cursor, and Codex. Use when bootstrapping lifecycle scripts, auditing a project's setup/run workflow, or wiring scripts into a runtime config like mise.toml or conductor.json.

Create semantic versioned releases with AI-generated changelogs. Worktree-aware - works from any branch. Use when the user wants to create a release, cut a release, bump version, or publish a new version. Supports dry-run preview, pre-releases (alpha/beta/rc), and CI status checks.

Session title generation, evaluation, and optimization. Auto-generates context-aware titles via Stop hook. Includes quality evaluation (pattern checks, LLM judge), GEPA-inspired prompt evolution, golden dataset tooling, and interactive title rating for training data collection.

Send structured logs to SigNoz observability platform. Use when you need to record events, errors, or activity for monitoring. Triggers on log, observe, signoz, telemetry, record event.

Guide for creating, editing, and evaluating skills. Use when creating a new skill, updating an existing skill, or verifying skills work before deployment. Also use when reviewing a third-party skill from skills.sh or GitHub, when asking "should I install this skill", "is this skill safe", "review this skill", "evaluate skill quality", or when comparing competing skills.

EXPERIMENTAL — only use when explicitly requested. Generate Claude Code skills from docs sites, GitHub repos, or local codebases using Skill Seekers.

Use when the user wants to list, search, install, remove, inspect, validate, audit, or update skills. Use when asking "what skills do I have", "is there a skill for X", "check my skills for issues", or "install a skill". Also use when checking skill health across agents (Claude Code, Codex, Agents CLI).

Customize or troubleshoot the live status line. Use when modifying statusline.sh, debugging token display, or understanding the rendering pipeline. The status line runs passively -- this skill is for working on it, not using it.

Write, review, or improve SwiftUI code following best practices for state management, view composition, performance, modern APIs, Swift concurrency, and iOS 26+ Liquid Glass adoption. Use when building new SwiftUI features, refactoring existing views, reviewing code quality, or adopting modern SwiftUI patterns.

Run deterministic GUI workflows in isolated Tart macOS VMs. Provides VM lifecycle management, guest command execution via tart exec (SSH fallback), VNC interaction (clicks, keyboard, captures), batch operations, target manifests for project-specific landmarks, and clean teardown. Headless by default with optional VNC observation.

Persistent AI teammate memory framework. Gives Claude Code agents persistent memory, evolving personality, and teammate-like behavior across sessions. Use when setting up personality.md injection, durable memory blocks in ~/.ai-memory, session start/end memory workflows, journal/remember/consolidate scripts, or packaging a distributable memory skill profile.

Smart dependency updates across ecosystems (npm/bun/pnpm, uv/poetry, cargo). Use when upgrading dependencies, fixing vulnerabilities, or performing proactive maintenance. Supports intelligent batching, risk assessment, and outcome tracking for continuous improvement.

Speak text aloud (TTS) and transcribe speech (STT). Supports local (macOS say, mlx-whisper) and cloud (ElevenLabs) providers. Use when user asks to speak, read aloud, listen, transcribe, or use vocal.

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

Interact with and test web applications using Playwright. Supports ad-hoc browser tasks (screenshots, form filling, deploy checks), Python and TypeScript E2E testing, visual debugging, and browser automation.

Extract and analyze YouTube video content (transcripts + metadata). Use when the user explicitly requests to analyze, summarize, extract wisdom from, or get context from a YouTube video. Supports wisdom extraction, summary, Q&A prep, key quotes, and custom analysis. Does NOT auto-trigger on YouTube URLs - only when analysis is explicitly requested.