runx
Search runx skills and inspect their manifests, requested authority and trust signals through the hosted read-only MCP server. Run skills locally with runx.
Hosted MCP Server
npx add-mcp 'https://api.runx.ai/mcp'Installs into Claude Code, Codex, Cursor and more
Documentation
runx agent guide
Governed runtime for agent skills: discover and install portable skills, run bounded skill graphs with explicit authority, and inspect signed receipts for what happened.
Machine surfaces
- SKILL.md: the canonical agent contract; start there.
- llms.txt: the machine surfaces and the official skill catalog as a plain-text index.
- llms-full.txt: the same index with every skill contract inlined.
- MCP server: streamable HTTP, no auth, read-only.
search_skillsfinds a skill,get_skillreturns its manifest, requested authority, trust signals and install command in one call. Manifest at https://api.runx.ai/mcp.json. - Documentation: hosted product and API docs.
- openapi.json: the hosted API contract, also served at https://api.runx.ai/openapi.json.
- Skill catalog: every published skill with per-skill pages at https://runx.ai/x/@/.
- Spec: the runx contract; schemas at https://schemas.runx.ai.
- ARD catalog: these surfaces as agentic resource discovery entries.
Use runx
Install the CLI from https://runx.ai/install, then follow https://runx.ai/SKILL.md: discover a skill in the catalog, run it under explicit bounded authority, and inspect the signed receipt the run leaves behind. https://runx.ai/start walks the same path for humans.