# Lrsi

> Lrsi is an open-source MCP server for agent tooling. Local MCP server for governed, audited self-improvement loops on your own machine. It is available on CLI, API and self-hosted.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, API, self-hosted
- **Repository:** https://github.com/anamized/lrsi
- **MCP Registry name:** io.github.ANAMIZED/lrsi
- **Categories:** [agent tooling](https://ai-powered.com/categories/agent-tooling)
- **Listing updated:** 2026-11-18

## About

The README describes LRSI as a local, closed-loop agentic operating system where every change to the system goes through a fail-closed boundary and an append-only event log. Changes are gated on evidence before they are promoted, so a mutation that fails its checks does not take effect.

Beyond the MCP server, the same core is exposed through a CLI, a Python SDK, a REST API and multi-agent workflows in which an improver, an evaluator and a council take turns. The project is built on the LRSI Runtime Core from a separate repository, which the README credits as its base.

The README is aimed at developers who want an auditable, self-hosted environment for agents that modify their own code or configuration. It installs from source with pip and includes a verify script and CI workflow.

## Features

- **Fail-closed mutation boundary:** Every change to the system is forced through a boundary that rejects it by default if governance checks do not pass.
- **Append-only audit trail:** Actions are recorded as events in an append-only log so past mutations can be reviewed.
- **Evidence-gated promotion:** Changes are promoted only after supporting evidence is collected, rather than automatically.
- **Multiple access surfaces:** The same functionality is available via CLI commands, a Python SDK, a REST API and an MCP server.
- **Multi-agent improvement workflow:** An improver, evaluator and council agent pass work between them in a defined sequence.

## Good for

- Running a self-hosted agent that proposes and tests changes to its own configuration
- Auditing which automated mutations were applied and why
- Exposing governed self-improvement tools to an MCP-compatible AI client
- Scripting improvement cycles through the Python SDK or CLI

## FAQ

### What license is LRSI released under?

The README lists the license as Apache-2.0.

### How do I install it?

The README gives the command pip install -e ".[dev,mcp,api]" from the repository, followed by bash scripts/verify.sh.

### Which entry point starts the MCP server?

The README lists lrsi-mcp as the MCP server entry.

### Is Lrsi free?

Yes. Lrsi is open source, so you can use and self-host it for free.

### What are the best alternatives to Lrsi?

The closest alternatives to Lrsi are Server Os, mcp-agent-template, RSOS State Runtime, librarian and Synod Council.

## Where this comes from

- Description: written from the project's README and its MCP Registry entry.

## Alternatives

- [Server Os](https://ai-powered.com/mcp/server-os): Server Os is an open-source MCP server for agent tooling.
- [mcp-agent-template](https://ai-powered.com/mcp/mcp-agent-template): mcp-agent-template is an open-source MCP server for security.
- [RSOS State Runtime](https://ai-powered.com/mcp/rsos-state-runtime): RSOS State Runtime is an open-source MCP server for agent tooling.
- [librarian](https://ai-powered.com/mcp/librarian-2): librarian is an open-source MCP server for agent tooling.
- [Synod Council](https://ai-powered.com/mcp/synod-council): Synod Council is an open-source MCP server for agent tooling.
- [Superagenticmcp](https://ai-powered.com/mcp/superagenticmcp): Superagenticmcp is an open-source MCP server for agent tooling.

---
Source: https://ai-powered.com/mcp/lrsi
