Enables discovery and installation of agent skills from curated GitHub repositories, providing tools to search collections, inspect skill contents, and download skills for local use.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Agent Skills MCP search for Rust development skills and install a useful one"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
pinkpixel-agentskills-mcp is a stdio MCP server for discovering, reading, and downloading agent skills from curated GitHub repositories.
GitHub: https://github.com/pinkpixel-dev/agentskills-mcp
It is built for a practical workflow:
search a large curated skill collection instead of searching all of GitHub
inspect a matching skill directly from GitHub
install a skill locally when the agent should actually use it
suggest a grounded starter scaffold when there is not an exact match
This server exists because a large skill library is only useful if an agent can actually find the right skill quickly.
With more than 1,600 collected skills spread across curated repositories, manual browsing becomes slow and noisy. This MCP server gives agents a direct way to search those collections, inspect likely matches, and install the right skill when it is needed.
Skills are genuinely useful when they are easy to discover and apply in context. The goal here is to make a large curated skill archive feel usable instead of overwhelming.
Related MCP server: skillet
github_skills_list_repositories
github_skills_search_skills
github_skills_get_skill
github_skills_install_skill
github_skills_suggest_skill_scaffold
Example user request:
Can you use the pinkpixel-agentskills-mcp tools and find skills for Rust development?Example result:
The server searches the built-in skill indexes.
It can identify strong matches like skills-collection-2:rust-pro and skills-collection-2:rust-async-patterns.
It can inspect those skill folders directly from GitHub before recommending them.
It can then install the selected skill locally with the MCP install tool.
This is especially helpful when a broad keyword search would otherwise return noisy matches, such as rust appearing inside trust.
Run from PyPI with uvx:
uvx pinkpixel-agentskills-mcpIf your environment still prefers the explicit package-to-command form, this works too:
uvx --from pinkpixel-agentskills-mcp agentskills-mcpRegister it in Claude:
claude mcp add github-skills -- uvx pinkpixel-agentskills-mcpWith a GitHub token for better rate limits:
claude mcp add github-skills --env GITHUB_TOKEN=$GITHUB_TOKEN -- uvx pinkpixel-agentskills-mcpThe server ships with these built-in default sources:
pinkpixel-dev/skills-collection-1
pinkpixel-dev/skills-collection-2
That means the server works out of the box with no repos.json at all.
Users can add more repositories in either of these ways:
Create repos.json in the project root by copying repos.example.json
Or set GITHUB_SKILLS_REPOS to a JSON array with the same schema
Each repo entry supports:
name: short alias used in skill slugs
owner: GitHub owner or org
repo: GitHub repo name
ref: branch or tag to read from
root: optional subdirectory that contains skills
github_token_env: optional environment variable holding a GitHub token
For public repositories, a GitHub token is optional. Users can run anonymously, or provide their own GITHUB_TOKEN for higher rate limits.
For private repositories, each user should provide their own token with the access they need. Do not ship your personal token with the server.
By default, custom repos are added on top of the built-in two repos.
If a custom repo uses the same name as a built-in repo, the custom one wins.
To disable the built-in repos entirely, set GITHUB_SKILLS_REPLACE_DEFAULTS=true.
To disable built-in repos without replacement, set GITHUB_SKILLS_INCLUDE_DEFAULTS=false.
For local development:
uv syncIf a user wants to add more sources, they can create repos.json from the example:
cp repos.example.json repos.jsonThis is a stdio server. To run it locally from the repo:
uv run agentskills-mcpFor a quick smoke test without leaving a hanging process:
timeout 5s uv run agentskills-mcpclaude mcp add github-skills --env GITHUB_TOKEN=$GITHUB_TOKEN -- uv run agentskills-mcpFor public repos, users can also add the server without any token:
claude mcp add github-skills -- uv run agentskills-mcpIf you also want a default install target for downloaded skills:
claude mcp add github-skills \
--env GITHUB_TOKEN=$GITHUB_TOKEN \
--env GITHUB_SKILLS_INSTALL_ROOT=/absolute/path/to/skills \
-- uv run agentskills-mcpThe published package name is pinkpixel-agentskills-mcp.
The server command is available as both:
pinkpixel-agentskills-mcp
agentskills-mcp
That means the most convenient public install path is:
uvx pinkpixel-agentskills-mcpIf you ever hit an environment that does not pick the matching executable automatically, use:
uvx --from pinkpixel-agentskills-mcp agentskills-mcpFor release steps, see PUBLISHING.md.
This server uses stdio, not HTTP/SSE transport.
Skill discovery is currently based on finding SKILL.md files in configured repos.
Built-in defaults make the server usable immediately, while optional config lets users extend the source list.
Search ranking is intentionally simple for the first version and can be upgraded later with repo-specific metadata or embeddings.
The scaffold tool is meant to help another agent create a new skill grounded in existing examples; it does not replace a full generation pipeline by itself.
Public-repo access works without credentials; tokens are an optional per-user enhancement, not a baked-in server secret.
Fetch the key files for a specific skill directory from GitHub.
Inputs: params (required). Returns: result
Download a skill directory from GitHub into a local folder.
Inputs: params (required). Returns: result
List the GitHub skill repositories configured for this MCP server.
Inputs: params (required). Returns: result
Search skill directories by matching the query against skill names and paths.
Inputs: params (required). Returns: result
Create a grounded starter scaffold for a new skill using the closest existing skills as examples.
Inputs: params (required). Returns: result
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool has a clearly distinct purpose: get_skill retrieves files, install_skill downloads to local, list_repositories shows configured repos, search_skills finds matches by query, and suggest_skill_scaffold creates new skill templates. There is no overlap in functionality, making tool selection straightforward for an agent.
All tools follow a consistent 'github_skills_verb_noun' pattern, using snake_case throughout. The verbs (get, install, list, search, suggest) are distinct and appropriate for their actions, creating a predictable and readable naming scheme across the toolset.
With 5 tools, this server is well-scoped for managing GitHub-based skills. Each tool serves a unique and necessary function in the skill lifecycle, from discovery to installation and creation, without being overly sparse or bloated.
The toolset provides complete coverage for the domain of GitHub skill management: list and search for discovery, get and install for retrieval, and suggest for creation. There are no obvious gaps, enabling agents to handle the full workflow from finding to implementing skills.
Search your team's shared AI-skill library, get install commands, and save skills from your agent.
Find, inspect and install reviewed reusable AI-agent capabilities from superskill.sh.
Search, fetch, lint, and install Agent Skills (SKILL.md) from the SkillMD registry.
Find AI agent skills: public previews and source links. Read-only discovery; no account required.
