Awesome Agent Skills AI Context Pack for Claude Code, Codex, Cursor, and Aider

Pack version: v1.0.0 · Last updated: 2026-05-14

Languages: English | 中文

---

This is an independent AI context resource pack for heilcheng/awesome-agent-skills: host instructions, prompt preview, evals, pitfalls, and recovery rules you can load into Claude Code, Codex, Cursor, Aider, and other AI coding agents.

The Problem

You found an awesome-agent-skills directory with 100+ skills. But you do not know which skill is safe to load into your AI host — some modify your AI config, some require API keys, and at least two have open security discussions. The repo name does not match the install command. No quick start tells you how to verify before committing.

This is an independent capability pack. It is not affiliated with or endorsed by heilcheng/awesome-agent-skills unless explicitly stated. It exists so your AI agent can answer those questions with evidence, not assumptions.

---

Copy / Run / Verify

1. Copy AGENTS.md (for Claude Code, Cursor, Codex, Gemini and other agent hosts) or copy CLAUDE.md (for Claude Code specifically) into your AI coding host.
2. Run the first prompt in 01_PROMPT_PREVIEW.md to experience the workflow before installing anything.
3. Verify with 06_EVALS/smoke_check.md before claiming any skill works.

Quick links: AGENTS.md · CLAUDE.md · Prompt Preview · Evals · Pitfalls · Manual · Boundary & Risk

---

AGENTS.md for Claude Code and AI Coding Agents

Use AGENTS.md for agent hosts that support repository-level instructions (Claude Code, Cursor, Codex, Gemini, etc.). Use CLAUDE.md when Claude Code is your target host and you want runtime-specific guidance including specific traps for awesome-agent-skills.

Both files include: when to load the pack, a recovery table for failed evals, a failure checklist, and explicit tool-use boundaries.

---

heilcheng/awesome-agent-skills Pitfalls and Recovery

Start with 03_PITFALL_LOG.md when setup, permissions, runtime behavior, or verification fails. The top 3 pitfalls are surfaced first with direct recovery steps and stop conditions. The full log remains navigable below.

---

What You Get

• Host instructions for AI coding agents (AGENTS.md / CLAUDE.md)
• A copyable prompt preview (01_PROMPT_PREVIEW.md)
• Smoke, boundary, and failure evals (06_EVALS/)
• Pitfall log with recovery steps (03_PITFALL_LOG.md) — top 3 are immediately actionable
• Boundary and risk card (04_BOUNDARY_RISK_CARD.md) — hard stops before any tool call
• Source attribution and upstream links (SOURCE_MAP.md)

If this pack helps your agent work from evidence instead of guesses, star the repo so future updates are easier to find. Open an issue for bugs, usage questions, or new pitfall reports.

---

Source Attribution

This pack was assembled by Doramagic to make heilcheng/awesome-agent-skills verifiable before adoption.

• Upstream: https://github.com/heilcheng/awesome-agent-skills
• Upstream license: MIT
• Upstream entry command: npx skills
• This pack adds: host instructions, evals, pitfall recovery, and boundary rules
• Relationship: independent pack; not affiliated with or endorsed by heilcheng/awesome-agent-skills unless explicitly stated

If you maintain the upstream project and want attribution changed or removed, open an issue in this repository.