
runx
Governed runtime for agent skills: search and inspect before you run
runx is an open-source MCP server for governance tooling. It runs as a hosted remote server, so there is nothing to install locally.
- Pricing
- Open source
- Platforms
- CLI, API, Self-hosted
- Runs
- Remotely
- Version
- 1.0.0
- Listing updated
Similar MCP servers
Install
Run in a terminal
claude mcp add --transport http runx https://api.runx.ai/mcpAdd to Settings → Connectors → Add custom connector
https://api.runx.ai/mcpAdd to ~/.cursor/mcp.json
{
"mcpServers": {
"runx": {
"url": "https://api.runx.ai/mcp"
}
}
}Add to .vscode/mcp.json
{
"servers": {
"runx": {
"type": "http",
"url": "https://api.runx.ai/mcp"
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"runx": {
"serverUrl": "https://api.runx.ai/mcp"
}
}
}Add to Grok Bot desktop app → Connect apps → Add a custom connector
https://api.runx.ai/mcpEnter a Name (for example "runx"), choose Type: Remote HTTPS, paste this as the Server URL and choose Add plugin.
Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"runx": {
"type": "http",
"url": "https://api.runx.ai/mcp"
}
}
}Using several servers? Build one config file with the MCP config generator.
About runx
runx treats a skill as a portable SKILL.md file: an operating manual with the method, the evidence to look for and the tools a job needs. Skills can call other skills, so a product launch or a market study can unfold as a chain of steps.
The runtime admits each call under explicit authority, hands credentials over only at the tool boundary and seals a signed receipt of what happened. Your own agent supplies the judgment while runx owns the execution boundary.
Through this MCP server an agent can search the skill catalogue and inspect a skill before running it. The runtime is written in Rust under Apache-2.0, with execution and receipt signing running locally and hosted registry services optional.
What it does
- Skills as files
- Each skill is a SKILL.md describing method, evidence, decisions and required tools.
- Catalogue search
- Agents search the skill catalogue and inspect a skill before executing it.
- Narrowing authority
- Calls are admitted under explicit authority that narrows rather than passes through.
- Signed receipts
- Every act produces a verifiable record sealed by the runtime.
- Credentials at the boundary
- Secrets are delivered at the tool boundary, not handed to the agent.
- Local execution
- Execution, state and receipt signing run on your machine; hosted services are optional.
Good for
- Run a product launch across docs, release notes and customer updates
- Investigate organic traffic with Search Console and analytics skills
- Audit what an agent did with signed receipts
- Share reusable expert procedures between agents
Pricing
Open source
Apache-2.0 runtime; hosted registry optional
Prices change. Check runx.ai before you buy.
runx FAQ
Is runx open source?
Yes, the Rust runtime is licensed under Apache-2.0.
Does runx run locally?
Execution, state and receipt signing run locally; the hosted registry, credential custody and notary services are optional.
How can I use runx?
Through the CLI, MCP or a language client alongside the agent or framework you already use.
Is runx free?
Yes. runx is open source, so you can use and self-host it for free.
How do I add runx to Claude Code?
Run this command in your terminal: claude mcp add --transport http runx https://api.runx.ai/mcp
What are the best alternatives to runx?
The closest alternatives to runx are Rokha, SkillGild and LoopSkill.
Where this comes from
- Description: written from the project's README, the project's website and its MCP Registry entry.
- Install steps: from the official MCP Registry entry.