Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/staruhub/claudeskills/agents-mdgit clone --depth 1 https://github.com/staruhub/ClaudeSkillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/staruhub/claudeskills/agents-md)<a href="https://agentmods.dev/instructions/staruhub/claudeskills/agents-md"><img src="https://agentmods.dev/badge/instructions/staruhub/claudeskills/agents-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00971 | $0.00971 |
| Opus 5 | $0.00485 | $0.00485 |
| Sonnet 5 | $0.00194 | $0.00194 |
| Haiku 4.5 | $0.00097 | $0.00097 |
Grade A, and why
ClaudeSkills AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Guidelines
Project Structure & Module Organization
Skills are stored as unpacked directories under skills/:
skills/
├── Geek-skills-xxx/
│ ├── SKILL.md # Core skill definition (source of truth)
│ ├── scripts/ # Optional executable scripts
│ ├── references/ # Optional reference documents
│ └── assets/ # Optional templates and resources
├── Geek-skills-yyy/
│ └── ...
Curated skills live under skills/ and must pass the L1 gates. Experimental or personal-use skills live under lab/ (same directory layout, Geek-skills-xxx naming) — they are excluded from scripts/validate.py, scripts/run_routing_evals.py, and scripts/install_skill.py, and are not listed in the curated index on the homepage. Promote a lab skill by moving its directory into skills/ and bringing it up to the quality standard; the reverse move demotes it.
Other top-level files:
README.md: project overview and curated skill index (English);README.zh-CN.mdis the Chinese version — keep both in synclab/README.md: index of experimental skillsAGENTS.md: repository guidelines (this file)
Prefer iterating on existing skills instead of introducing new parallel variants.
Build, Test, and Development Commands
Most changes are content edits to skill files. Two canonical checks exist:
python3 scripts/validate.py— structural L1 assertions for every curated skill (frontmatter, line caps, 三件套 sections, orphan files, platform-path/CVE hardcoding). Run before publishing any skill change.python3 scripts/run_routing_evals.py— validates allevals/routing-evals.jsonfiles (schema, global id uniqueness, route_to targets, cross-skill prompt conflicts). Add--emit-promptsto produce an agent-based L2 routing test bundle. Run after changing any skilldescription.python3 scripts/install_skill.py <name>— install a skill under a clean command name (--list,--project,--force,--dry-run).
CI (.github/workflows/validate.yml) runs the two L1 gates plus a script compile-check on every push and PR.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 64 lines · 971 tokens per session scan A 80b6f40c6b9c
ClaudeSkills AGENTS.md is an instructions file published in the GitHub repository staruhub/ClaudeSkills (707 stars, last pushed 23d ago), licensed MIT. It adds 971 tokens to every session, about $0.0049 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
foundry-research CLAUDE.md
Instructions for foundry-works/foundry-research, covering foundry research, plugin development, plugin gotchas and design principles.
SeekMoney-ai CLAUDE.md
Claude Code instructions for zykooooooooo/SeekMoney-ai, covering claude.md, project overview, development commands, development and production.
critical-thinking AGENTS.md
Instructions for Johna2an/critical-thinking, covering agents.md, repo map, install (for a user asking "should i install this?"), working in this repo and running an eval round.
hometax-doum AGENTS.md
AGENTS.md instructions for lbiz-partners/hometax-doum, covering claude.md — hometax-doum 개발 규칙 (하네스와 쌍), 규칙 ↔ 강제 장치 and 운용 메모.
powerglide AGENTS.md
Instructions for bkataru/powerglide, covering powerglide agent protocol & integration guide, the ralph loop protocol, termination signals, velocity control and reliable pty execution.
chealth CLAUDE.md
Instructions for danielithomas/chealth, covering claude.md, project purpose, language & tooling, build & test and architecture.