skills

Agent Skills

Open-source Agent Skills by David Ondrej, for Claude Code, Codex, Pi, and any agent that loads SKILL.md files.

Run them on cloud agents with Cloudroom, free and open source.

Install

Requires Node.js. Run from your project, or add -g to install globally:

npx skills add davidondrej/skills                       # pick from all skills
npx skills add davidondrej/skills --skill deepapi -a claude-code

Or copy any skills/<name>/ folder into your agent's skills directory.

Layout

skills/<name>/SKILL.md     instructions, with name + description frontmatter
skills/<name>/references/  optional docs the agent loads on demand
skills/<name>/scripts/     optional helper scripts
AGENTS.md                  David's global agent instructions

Skills

• 4x4 — Pull 16 fast decisions out of the user with 4 sets of 4 open questions, then present an implementation plan.
• adr-verbatim — Record concise ADRs that preserve the user's decisions without conversational noise.
• anti-sleep — Keep a Mac awake for a set duration or while a process runs.
• ask-then-build — Clarify a feature, change, or refactor through questions, then write a build prompt for another agent.
• bb-cli — Use this when controlling bb.
• bb-plugins — Build, install, and debug bb plugins.
• before-building — Fire the moment the user proposes a build.
• better-prompt — Rewrite a prompt into a deeper version of the same task, the way a top professional would approach it.
• boat — Manage Boat (boat.dev, formerly Box by Ascii) cloud sandboxes via API, CLI, and SSH.
• browser-harness — Direct browser control via CDP.
• browser-use — Direct browser control via CDP for web interaction: automation, scraping, testing, screenshots, and site/app work.
• cmux — Control the cmux macOS terminal app: workspaces, panes, surfaces, browsers, notifications, settings, and hooks.
• codex-subagent — Launch OpenAI Codex CLI as a subagent (ChatGPT subscription auth, no API key).
• cua-driver — Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording.
• decisions — Ask the agent to list all choices it made during the current work that it is not confident of.
• deep-research — Run deep research through DeepAPI and save a report with sources.
• deep-scrape — Build sourced JSON dossiers on people, companies, or topics with DeepAPI.
• deepapi — Use DeepAPI for all web search, deep research, and web scraping (websites, LinkedIn, GitHub, X/Twitter, YouTube, Instagram) instead of built-in search, research, fetch, or browser tools.
• designs — Show 4 better, new design options as separate real image renders.
• domain-checker — Check domain registration in batches using public RDAP APIs, without keys or login.
• effective-agent-skills — Write, review, and debug agent skills.
• fable-review — Launch an independent Fable code review and return it verbatim.
• fable-safe-prompt — Make minimal prompt edits to reduce false-positive refusals from Fable.
• file-tree — Manually-invoked skill that prints a nice file tree of the current topic in a code block.
• fireflies-transcript — Fetch raw Fireflies.ai meeting transcripts.
• gemini-review — Launch an independent Gemini 4 Carbon code review through Cursor at High reasoning and return it verbatim.
• git-worktree — Use Git worktrees to isolate parallel coding tasks.
• github-outside-sandbox — Run Git and GitHub CLI commands on the host when sandboxing blocks Keychain auth, network access, or .git writes.
• global-agent-guardrails — Configure the shared guard against catastrophic shell commands in local AI agents.
• goal-loop — Draft /goal prompts and explain persistent agent loops.
• gpt-review — Launch an independent GPT-6-Astra code review and return it verbatim.
• handoff — Capture session context for a fresh agent.
• herdr — Operate and coordinate AI agents in Herdr workspaces in Ghostty.
• image-prompt — Write four distinct prompts for AI image models.
• keep-track — Keep track of open topics and decisions.
• kimi-review — Launch an independent Kimi K3.1 Pro Max code review and return it verbatim, falling back across harnesses and routes until one works.
• larp-detector — Skeptical review of outside contributions to public repos.
• muse-review — Launch an independent Meta Muse Wonder Max code review through Pi and OpenRouter, and return it verbatim.
• nagent — Launch a new bb worker thread with the right project, model, worktree, and task brief.
• occams-razor — Stop overthinking and answer with what we actually know for sure, how you know, and a confidence %.
• openrouter — Design, build, debug, and optimize OpenRouter API integrations.
• persistent-localhost — Manage persistent dev servers, APIs, and other local processes on a port using macOS LaunchAgents.
• prompt-for-others — Write a prompt teammates can give their AI agents to apply a fix, upgrade, or setup change.
• read-all-adrs — Read every ADR markdown file in the project's docs/adr/ folder so you have full context on past decisions.
• remind — Rewrite the last response simpler and shorter in plain English, prefixed with a 3-5 sentence TLDR of the conversation so far.
• rename — Manually-invoked skill that renames the current thread or session to 2-5 lowercase descriptive words.
• rename-process — Coordinate project or repository renames and moves involving Git, BB, services, or multiple machines.
• repo-sync — Manage automatic Git syncing on macOS with repo-sync.
• risky-changes — Verify assumptions before implementing large or risky changes to APIs, provider data, billing, pricing, quotas, or defaults.
• self-archive — Archive the current Cloudroom thread and release its runtime.
• setup-help — Guide setup one step at a time and show what remains.
• state — Manually-invoked skill that recaps what the current thread is doing and how far along it is.
• stop-overthinking — Manually-invoked skill that stops overthinking and forces a short, practical decision.
• total-review — Run five independent max-effort reviews (Fable 5.1, Opus 5.5, GPT-6 Astra, GPT-6.1 Sol, Muse Wonder), merge their findings, and fix only what the user approves.
• varied-search — Research a topic across the web, GitHub, and X/Twitter with DeepAPI and produce a concise Markdown report.

Contributing

This repo is generated from a private source repo and overwritten on every publish, so pull requests can't be merged.

License

MIT