code-review
Review code for quality, security, and best practices. Use when reviewing pull requests or when the user asks for a code review. Claude, Codex, GitHub Copilot compatible.
★ 92 👁 12k
View & Install
AI Agent Skills, the skill dock for sharp agents.
Upload Agent Skills bundles, version them like npm, and make them searchable. No gatekeeping, just signal. Works with Claude, Codex, GitHub Copilot, VS Code, and more.
Search skills. Versioned, rollback-ready.
Install any skill folder in one shot:
npx aiagentskills@latest install code-reviewHighlighted skills
Curated signal — highlighted for quick trust.
commit-helper
Generate commit messages from git diffs. Use when writing commit messages. Follows conventional commits; works with any agent that supports SKILL.md.
★ 84 👁 9.2k
View & Install
pdf-processor
Extract text, tables, and metadata from PDFs. Use when working with PDF files or document extraction. Based on the open Agent Skills standard.
★ 64 👁 5.1k
View & Install
Popular skills
Most-downloaded, trusted picks.
readme-writer
Generate project READMEs from codebase structure and conventions. Use when user asks for documentation or README.
★ 1.2k · 18k
View & Install
test-fixing
Detect failing tests and propose fixes. Use when CI fails or user asks to fix tests.
★ 480 · 11k
View & Install
git-workflow
Automate git operations and repository interactions. Use for branching, rebasing, or release workflows.
★ 644 · 8.5k
View & Install
What Are Agent Skills?
Agent Skills are how-to guides for AI assistants. Instead of the AI needing to know everything upfront, skills let it learn new abilities on the fly — like giving someone a recipe card instead of making them memorize an entire cookbook. Skills are simple text files (SKILL.md) that teach an AI how to do specific tasks.
How it works (progressive disclosure)
- Browse — The AI sees a list of available skills (names and short descriptions only).
- Load — When a skill is needed, the AI reads the full instructions.
- Use — The AI follows the instructions and accesses any helper files or scripts.
This keeps context lean: you can install many skills without bloating the context window. Awesome Agent Skills · DeepLearning.AI course
Compatible agents
Skills follow an open standard (e.g. agentskills.io). Works with: Claude Code, Claude.ai, OpenAI Codex, GitHub Copilot, VS Code, Antigravity (Google), Kiro, Gemini CLI, and more.
Why AI Agent Skills
Agents orchestrate workflows; skills package reusable expertise. Use both: Agents or skills? The answer is both.
| Capability | Generic prompts | AI Agent Skills |
|---|---|---|
| Consistent behavior | Varies by session | Versioned, repeatable |
| Team standards | Manual copy-paste | Shared skills, one install |
| Context window | Everything in prompt | Progressive disclosure (metadata → full content → resources) |
| Fine-tuning | Required for new behavior | No fine-tuning; update SKILL.md |
| Rollback | Not available | Pin versions, rollback anytime |
Four core advantages
Specialize, reuse, and compose your AI capabilities. Manus and Spring AI use the same ideas.
Specialization
Customize AI for specific domains: legal review, financial analysis, code review, docs.
Reusability
Build once, use everywhere. Deploy skills across conversations, projects, and compatible platforms.
Composability
Combine multiple skills for complex, multi-step workflows — no fine-tuning required.
Standardization
SKILL.md is an open standard. Enforce team best practices and consistent execution.
Use Cases & Integrations
Ai skills and agent skills for code, docs, and automation. Agent skills Claude, openclaw agent, skills GitHub.
- AI agent – Code review, refactors, commit messages, test generation
- Agent skills Claude – Claude and Anthropic workflows
- Skills GitHub – PR checks, release notes, repo automation
- Openclaw agent – Openclaw-compatible agent skills
- IDEs & editors – Cursor, VS Code, CLI
Quick Start: Using AI Agent Skills
Get ai skills and agent skills running in minutes.
- Ensure Node.js is installed. Then run:
npx aiagentskills install code-review - The skill is installed into your project or global config (e.g.
~/.claude/skills/or.github/skills/). Follow the skill’s README for any env or config. - Use your AI agent in IDE or CLI; the skill is applied when the task matches its description (the description drives when the agent activates the skill — SKILL.md pattern).
What People Say About AI Agent Skills
Teams using ai skills and agent skills in production.
"Agent skills for our AI agent cut PR review time in half. Skills GitHub integration is smooth."
"Best ai skills we've found. Agent skills Claude and openclaw agent both work. One install."
"Ai skills let us encode our docs. Our AI agent actually follows them. Game changer."
Frequently Asked Questions
Common questions about ai agent, ai skills, and agent skills.
An AI agent is an autonomous or semi-autonomous system that uses AI to perform tasks. Agent skills extend what your AI agent can do with reusable, versioned instructions (e.g. SKILL.md).
Agent skills are reusable instruction sets (typically in SKILL.md) that teach your AI how to do specific tasks—code review, commit messages, PDF processing—with consistent, versioned behavior. They use progressive disclosure: only metadata is loaded at startup; full instructions load when the task matches. No fine-tuning required.
Agent Skills = teach the AI how to do something (workflows, best practices). MCP = help the AI access things (APIs, databases, external tools). They work together: skills define the playbook; MCP provides the data pipeline.
Fine-tuning permanently changes how a model behaves (expensive, hard to update). Agent skills are just instruction files—you can update, swap, or share them anytime without touching the model. Build once, use across Claude, Codex, Copilot, and more.
Agent skills Claude are skills designed to work with Claude (Anthropic). They follow the same SKILL.md format and can be used in Claude Code, Cursor, and other Claude-integrated tools.
Many agent skills are hosted on GitHub. You can install from the registry (one command) or from a GitHub repo URL. The registry indexes and versions skills for easy discovery—skills GitHub style.
Openclaw agent refers to the OpenClaw ecosystem for AI agents. AI Agent Skills offers openclaw agent–compatible skills so you can use the same SKILL.md format across OpenClaw and other agents.
Run npx aiagentskills install <skill-name> from your project or home directory. The CLI installs the skill and any documented dependencies. You can pin a version for repeatable installs.
Create a folder with a SKILL.md file (name, description, instructions). Add metadata (version, dependencies). Test locally, then run npx aiagentskills publish and follow the prompts to submit to the registry.
A SKILL.md file has a YAML frontmatter (name, description, optional metadata) and a markdown body that describes what the skill does, when to use it, and step-by-step instructions. Natural language is enough.
Re-run install with a specific version: npx aiagentskills install <skill-name>@<version>. Your config or lockfile can also pin versions for deterministic installs.
Check the documentation in the footer, the GitHub repo for issues and discussions, and the community Discord. For skill-specific help, open an issue on the skill’s repository if available.
Ready to use AI agent skills?
Browse the registry or publish your own ai skills and agent skills.