loopkit
33 battle-tested skills + minimal .claude harness for any coding agent (Claude Code, Cursor, Codex, Gemini CLI).
Install with your AI
Paste into Claude Code, Cursor, or any agent — it reads the repo and wires the tool into your project.
Install and set up loopkit (claude-plugin project) into my current project. Found on https://claudeers.com/loopkit Repo: https://github.com/Archive228/loopkit Homepage/docs: — Detected install method: claude-plugin → /plugin install loopkit@Archive228/loopkit Category: mcp-servers. Platforms: cli, api. Read the repo's README for exact setup and env vars, then install it and wire it into my project. Claudeers Health Verdict: active; community-verified: false. Confirm the source before running anything.
/plugin marketplace add Archive228/loopkit /plugin install loopkit@Archive228/loopkit
git clone https://github.com/Archive228/loopkit
// compatibility
| Platforms | cli, api |
|---|---|
| Operating systems | — |
| AI compatibility | claude |
| License | MIT |
| Pricing | open-source |
| Language | JavaScript |
loopkit
A curated library of 33 battle-tested skills that make any coding agent ship cleaner code — and stop guessing.
Drop them into .claude/skills/ (or any agent that reads SKILL.md) and your agent loads each one only when it's relevant: debug systematically, review for security, design real UIs, run agents in a lean loop, write the failing test first, ship clean commits.
curl -fsSL https://raw.githubusercontent.com/Archive228/loopkit/main/install.sh | bash
Compatible with Claude Code · Cursor · Codex · Gemini CLI. Built on the agent-engineering patterns Anthropic publishes (Building Effective Agents, Claude Code docs) — distilled into skills you can actually drop in.
What the installer writes
Into your current directory:
.claude/CLAUDE.md— 60-line standing context.claude/settings.json— permission allowlist + format-on-write hook.claude/agents/verifier.md— adversarial verifier subagent.claude/skills/— the 33 skill files (load only on trigger).mcp.json— MCP server wiringMEMORY.md— cross-session memory indexrun.sh— Plan→Act→Verify loop runner
Existing files are kept by default. Pass FORCE=1 to overwrite:
curl -fsSL https://raw.githubusercontent.com/Archive228/loopkit/main/install.sh | FORCE=1 bash
Or clone and copy manually:
git clone https://github.com/Archive228/loopkit
cp -r loopkit/.claude loopkit/skills loopkit/.mcp.json loopkit/MEMORY.md loopkit/run.sh your-project/
Why skills
A raw agent guesses. When it guesses, it fabricates — invented APIs, tests that pass nothing, "fixes" that rename the bug. A skill is a tight, reusable instruction set that loads only when its trigger matches, so the agent specializes instead of winging it — without paying for every skill's tokens on every prompt.
The library (33 skills, 9 tracks)
| track | skills |
|---|---|
| agent/llm | context-budget · spec-first · tool-restraint · subagent-fanout |
| debug | systematic-debugging · read-the-trace · bisect-regression |
| security | owasp-review · authz-check · input-validation · secret-scan · dependency-audit |
| frontend | design-system · a11y-pass · loading-empty-error-states |
| testing | write-failing-test-first · flaky-hunter · coverage-gaps · contract-test |
| refactor | kill-dead-code · simplify · reduce-nesting |
| docs | changelog-from-diff · decision-record · readme-audit |
| data | sql-review · migration-writer · schema-diff |
| git/ops | clean-commits · pr-from-diff · rebase-safely · revert-surgical |
| review | adversarial-verify (the 11 shortcuts agents take to fake "done") |
Each skill: a focused technique, not a tutorial. Read one, fork it, make it yours.
Bonus: the harness
init also drops a minimal working .claude/ (standing context, permission allowlist, format-on-write hook, a verifier subagent, MCP wiring, memory index) and a Plan→Act→Verify loop runner — the floor your skills run on. Full theory: Loop and Harness engineering: 7 files, 5 steps.
MIT. Built to be forked.
// faq
What is loopkit?
33 battle-tested skills + minimal .claude harness for any coding agent (Claude Code, Cursor, Codex, Gemini CLI).. It is open-source on GitHub.
Is loopkit free to use?
loopkit is open-source under the MIT license, so it is free to use.
What category does loopkit belong to?
loopkit is listed under mcp-servers in the Claudeers registry of Claude-compatible tools.
// embed badge
[](https://claudeers.com/loopkit)
// retro hit counter
[](https://claudeers.com/loopkit)
// reviews
// guestbook
// related in MCP Servers
🔓f.k.a. Awesome ChatGPT Prompts. Share, discover, and collect prompts from the community. Free and open source — self-host for your organization with complete…
🔓A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io
🔓An open-source AI agent that brings the power of Gemini directly into your terminal.
🔓🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman