Agent Toolkit

A curated collection of skills, commands, plugins, and workflows for AI agents. Framework-agnostic — works with any AI agent that can read a skill directory and execute shell scripts or code.

Every skill in this toolkit follows the AgentSkills.io standard. Each skill is a self-contained directory with a SKILL.md spec, versioned independently, and compatible with any agent framework that implements the AgentSkills specification.

What's in this toolkit

Type	Description
Skills	Self-contained units of capability the agent loads and invokes
Commands	Reusable shell or script commands the agent can run
Plugins	Extensions that add new tools or integrations to the agent
Workflows	Multi-step automation sequences combining skills and commands

Skills

All skills listed below are fully compliant with the AgentSkills.io specification.

Skill	Description
ai-trending-news	Fetch, rank, and summarize the top 10 trending AI news stories
github-trending	Fetch top trending GitHub repositories for a given time range
openclaw-backup-restore	Backup and restore a complete OpenClaw installation as a portable archive
openclaw-git-sync	Version ~/.openclaw with git and push to a remote repo, with auto-generated commit messages

AgentSkills.io standard

This toolkit strictly follows the AgentSkills.io open standard for agent skills. Every skill in this repository:

• Has a SKILL.md file with valid frontmatter (name, description, license, metadata)
• Is fully self-contained — no shared dependencies between skills
• Declares its compatibility and system requirements explicitly
• Works with any agent framework that implements the AgentSkills specification

Skill structure

skills/<skill-name>/
├── SKILL.md          — AgentSkills spec: metadata + instructions for the agent
├── README.md         — human-readable documentation
├── scripts/          — executable scripts the agent can invoke
└── references/       — supporting data, configs, or policy files

Compatibility

Because all skills follow the AgentSkills.io standard, they are compatible with any AI agent that supports the specification — not tied to a single framework.

Tested with: OpenClaw, Claude Code, Codex

Adding a skill

New skills must follow the AgentSkills.io standard:

1. Create a directory under skills/<your-skill-name>/
2. Add a SKILL.md with valid AgentSkills frontmatter (name, description, license, metadata) and clear instructions
3. Add scripts under scripts/ and any reference data under references/
4. Add a README.md for human readers
5. List the skill in this README

License

Apache-2.0