Agent Skill Runtime MCP Server
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 Skill Runtime MCP Serverlist the available skills, then run example_run with message hello"
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.
Agent Skill Runtime
@lunarmoon26/agent-skill-runtime turns skill-owned JSON-in/JSON-out scripts into tools for MCP, OpenCode, and DeepSeek Harness. A versioned skill-runtime.json file supplies the portable tool description; thin host adapters preserve each harness's native lifecycle and cancellation behavior.
Status
This repository contains the version 1 contract and its reference TypeScript
implementation. Version 0.1.1 is the current pre-release.
Related MCP server: Skillz
Quick Start
Place a manifest beside a skill's SKILL.md:
{
"$schema": "https://raw.githubusercontent.com/lunarmoon26/agent-skill-runtime/main/skill-runtime.schema.json",
"manifestVersion": 1,
"name": "example-skill",
"tools": [
{
"title": "Run Example",
"name": "example_run",
"description": "Run the example workflow.",
"inputSchema": {
"type": "object",
"properties": {
"message": { "type": "string" }
},
"required": ["message"],
"additionalProperties": false
},
"entrypoint": {
"engine": "node",
"path": "scripts/main.mjs"
}
}
]
}The entrypoint reads one JSON value from stdin, writes one JSON value followed by a newline to stdout, writes diagnostics to stderr, and exits nonzero on failure.
npx @lunarmoon26/agent-skill-runtime validate --root .
npx @lunarmoon26/agent-skill-runtime list --root .
printf '{"message":"hello"}' | npx @lunarmoon26/agent-skill-runtime run example_run --root .
npx @lunarmoon26/agent-skill-runtime mcp --root .python-uv entrypoints declare exact PEP 723 dependencies in their script. Run prepare before invoking them; normal tool execution uses uv --offline and never downloads packages.
Package Exports
@lunarmoon26/agent-skill-runtime: discovery, validation, preparation, and execution.@lunarmoon26/agent-skill-runtime/opencode: OpenCode tool adapter.@lunarmoon26/agent-skill-runtime/deepseek: DeepSeek Harness tool adapter.@lunarmoon26/agent-skill-runtime/mcp: MCP stdio server.
See docs/contract.md for the execution and security rules and docs/integrations.md for host examples.
Development
npm install
npm test
npm run verifyPull requests and pushes to main run the full verification suite on Node.js
22.20 and 24, plus the runtime tests on Bun 1.4.
Releases
npm releases use trusted publishing
from GitHub Actions. The release workflow accepts only a vX.Y.Z tag that
matches the committed package.json version and points to a commit contained
in main. It repeats the Node and Bun verification before publishing. npm uses
the workflow's short-lived OIDC identity and adds provenance automatically; no
long-lived npm token is stored in GitHub.
Configure the npm package's GitHub Actions trusted publisher once with these exact values:
Organization or user:
lunarmoon26Repository:
agent-skill-runtimeWorkflow filename:
npm-publish.ymlEnvironment: none
Allowed action: direct
npm publish
For each release:
Update
package.jsonandpackage-lock.jsonto the same new version in a reviewed change, then merge it tomain.Tag the merged
maincommit and push the tag:git tag vX.Y.Z git push origin vX.Y.Z
The tag starts .github/workflows/npm-publish.yml. Branch pushes and pull
requests never publish. After the first trusted release succeeds, configure
the npm package to require two-factor authentication and disallow traditional
publish tokens.
Version 0.1.0 was published manually before release automation existed. Do
not create a retrospective v0.1.0 tag after this workflow reaches main; the
first automated release must use a new package version and matching tag.
This server cannot be deployed
Maintenance
Related MCP Connectors
The governed runtime for agent skills. Search the catalog and inspect a skill before running it.
Search and discover Agent Skills from the skills.sh registry. Powered by HAPI MCP server.
Search, fetch, lint, and install Agent Skills (SKILL.md) from the SkillMD registry.
Governed AI agent skills — one library, distributed to devs and exposed to remote agents over MCP.
Related MCP Servers
- AlicenseAqualityCmaintenanceConverts AI Skills (following Claude Skills format) into MCP server resources, enabling LLM applications to discover, access, and utilize self-contained skill directories through the Model Context Protocol. Provides tools to list available skills, retrieve skill details and content, and read supporting files with security protections.328Apache 2.0
- AlicenseNot gradedqualityDmaintenanceTurns Claude-style skills (SKILL.md files with resources) into callable MCP tools for any agent. Discovers skills from a directory, exposes their instructions and resources, and can execute bundled helper scripts.162 PyPI401MIT
- AlicenseNot gradedqualityNot gradedmaintenanceAn MCP server that transforms Claude-style skills and resources into callable tools for any MCP-compatible agent or client. It automatically discovers, exposes, and executes scripts from skills organized in local directories or packaged archives.MIT
- AlicenseAqualityAmaintenanceMCP server that enables running and managing file-based AI Skills, Agents, and Flow workflows as DAGs, with tools for listing, executing, and resuming tasks via stdio JSON-RPC.9MIT