Thank you for your interest in improving and growing this open-source agent skills gallery!
We curate skills that:
- Solve real, complex engineering problems (not generic prompts like "write an email").
- Offer technical depth (edge cases, pitfalls, concrete commands, scripts, and error diagnostics).
- Support multiple AI agent runtimes (Hermes, Claude, Cursor, Gemini, GPT).
Every skill must live in a subdirectory under its relevant category (e.g. skills/agent-ops/<skill-name>/) and contain:
README.md: Human-readable guide explaining the problem, prerequisites, architecture, and examples.SKILL.md: Hermes Agent specification with YAML frontmatter (name,description,version,tags).CLAUDE.md: Claude Code / Anthropic Projects prompt.CURSOR.mdc: Cursor Rules format (description,globs, guidelines).GEMINI.md: Google Gemini system instruction format.GPT.md: OpenAI Custom GPT instruction format.
Before opening a Pull Request, ensure:
- ZERO Credentials: No real API keys, bearer tokens, passwords, or hashes.
- Sanitized Variables: Use clear placeholders (e.g.,
YOUR_TELEGRAM_BOT_TOKEN,YOUR_RADIO_ID,example.com). - No Private Domain Names / IPs: Use generic domains or private RFC1918 ranges (
192.168.1.x,10.0.0.x). - Empirical Verification: Commands and scripts must have been tested in a real terminal environment.
- Fork the repository.
- Create a feature branch:
git checkout -b feat/new-skill-lora-tuning
- Add your files following the template structure.
- Commit and Push:
git commit -m "feat(skills): add LoRA tuning automation skill" git push origin feat/new-skill-lora-tuning - Open a Pull Request describing the motivation, tested environments, and use cases.