What skills format does it expect?
Each skill is a folder containing a `SKILL.md` file.
@back1ply
Basic information
MCP server to expose Claude Code Skills to AI agents
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"agent-skill-loader": {
"command": "npx",
"args": [
"-y",
"agent-skill-loader"
]
}
}
}Tool list not available yet
This listing does not currently have a tool list to display. This does not mean the server has no tools; check its documentation for supported capabilities.
Fetch the live tool list by running this server in a temporary sandbox using the button above.
Agent Skill Loader is a Model Context Protocol (MCP) server that bridges a static Claude Code Skills library with dynamic AI agents (like Antigravity, Claude Desktop, or Cursor). It allows agents to learn skills on demand without requiring manual file copying into every project.
Install globally via npm (npm install -g agent-skill-loader) and register the server in .mcp.json. Alternatively, build from source. After setup, agents have access to five tools: list_skills, read_skill, install_skill, manage_search_paths, and debug_info.
Each skill is a folder containing a `SKILL.md` file.
Use the `manage_search_paths` tool at runtime with `operation="add"` and the path.
Yes, use `install_skill(skill_name)` which copies the folder to `.agent/skills/<name>` within the current workspace.
Node.js >= 18.
It scans `%USERPROFILE%\.claude\plugins\cache` and a `skill-paths.json` file in the project root.

Social listening for AI agents. Rumoro's remote MCP server lets Claude, Cursor, Codex and other MCP clients search and sort mentions of your product, your competitors and your topics on Reddit, X, Hacker News, GitHub, Yo

Discover flat-rate DeepSeek plans, model limits, API-key instructions, and coding-client setup
TruVerifAI's Panel Review is the guardian agent for AI's highest-stakes coding decisions: four frontier models argue over an AI coding agent's riskiest designs, diffs, and commits before they execute, with local review g
Model Context Protocol (MCP) server for Hostinger API. If you don't want to install or run anything locally, connect directly to Hostinger's hosted MCP server: This opens a browser window to authorize via OAuth. Once app

Singularis is unlimited computation. Bring us your hardest problems. MCP server; free agent key: 500 credits instantly.
Comments