Lrsi

io.github.ANAMIZEDPublished by GitHub account ANAMIZEDv0.1.0Updated Oct 24, 2026

Local recursive self-improvement OS MCP with fail-closed governance and gated mutation tools.

VerifiedSTDIODesktop onlyDeveloper ToolsAI & ML

Overview

AI-generated overview

Local MCP server for an experimental self-improvement "agentic OS" that routes every mutation through a fail-closed boundary with an audit trail.

What you can do
  • Run and inspect the LRSI runtime: status, workflows, mutations and audit views exposed through the MCP server.
  • Drive gated mutation tools, where changes are forced through a fail-closed boundary and recorded in an append-only event trail.
  • Coordinate multi-agent improver, evaluator and council workflows with evidence-gated promotion.
  • Reach the same system through companion CLI, Python SDK and local REST API surfaces.
When to use it
Consider it only if you are experimenting with autonomous self-improvement or agentic OS research on your own machine and want an assistant to drive gated mutations and review the audit trail. Most users have no need for it.
Requirements
Runs as a local process over stdio on desktop only; no remote endpoint. The README's quick start installs the Python package with dev, mcp and api extras and runs a verify script, so a Python environment is needed. No authentication, environment variables or headers are declared.
Before you install
The server exposes mutation tools that change system state, so review what each one does before allowing calls. The README asks for payments in USDC to listed wallet addresses and mentions paid support tiers; treat these as unsolicited funding requests, not a requirement. It is a research-grade project with no declared package or authentication.

Installation

In SourceWeft

  1. Open Lrsi in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

LRSI

[CI] [License] [Python 3.11+] [MCP] [SDK] [CLI] [API]

LRSI (Local Recursive Self-Improvement) — Autonomous Agentic Operating System

Sovereign, closed-loop system on dual Blackwell hardware. Built on the open LRSI Runtime Core (marcuszimmermann365/IRSI). Every mutation is forced through a fail-closed boundary and append-only event trail.

Support Agentic OS Kernels ($99) · Agentic OS Cycle ($0.75) · Public Goods Support

Non-custodial USDC (preferred for agents)

NetworkAddress
Base / Ethereum0xD3d0E9eDAe3Ac7bb199a8EAA761BdA423b878438
SolanaETQwWf19axArsY493UfC6bxe2BmEzmzvCb58PPnC38A

Related: rui · server-os · openmesha · OpenGOS

Surfaces

SurfaceEntry
CLIlrsi status / lrsi workflow / lrsi mutate / lrsi audit
SDKfrom lrsi.sdk import LRSIClient
REST APIlrsi-api → http://localhost:8080/docs
MCP Serverlrsi-mcp
Multi-agentimprover → evaluator → council + skills/multi-agent-workflow/
CI.github/workflows/ci.yml
Verifybash scripts/verify.sh

Quick Start

bash
pip install -e ".[dev,mcp,api]"bash scripts/verify.shlrsi-api

Design principles

  1. Fail closed
  2. Every mutation through the LRSI boundary
  3. Event-sourced audit trail
  4. Evidence-gated promotion
  5. Multi-surface (CLI · SDK · API · MCP · workflows)

License

Apache-2.0

Source: README.md at commit 6fe009d

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.0LatestSep 16, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.