Skillget

libr-agent

mcp
Security Audit
Warn
Health Warn
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Low visibility — Only 7 GitHub stars
Code Pass
  • Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
LibrAgent is a local-first desktop application that provides a persistent, stateful workspace for running autonomous AI agents. It allows agents to continuously interact with a unified terminal and live browser sessions while orchestrating complex tasks using the Model Context Protocol (MCP).

Security Assessment
This tool executes shell commands, automates web browsers, and makes external network requests to interact with third-party AI models. A rule-based scan of the codebase found no hardcoded secrets or dangerous patterns, and the application does not request elevated system permissions. However, because it features a "YOLO Mode" for autonomous execution of sensitive tools without manual approval, the potential for unintended local actions is significant if an agent behaves unpredictably. Overall risk is rated as Medium.

Quality Assessment
The project is licensed under the permissive MIT license and is under active development, with its most recent code push happening today. The code is written in Rust and is built on the reputable Tauri framework, which provides a secure foundation for desktop applications. However, it currently has very low community visibility with only 7 GitHub stars, meaning it has not been broadly tested or audited by a wide user base. Trust should be evaluated with caution due to its early stage of public adoption.

Verdict
Use with caution: the underlying code appears safe, but users should carefully manage the autonomous execution features given the tool's early stage and low community visibility.
SUMMARY

Desktop AI agent with built-in tools. No complex setup, just productivity.

README.md

🤖 LibrAgent

A lightweight, stateful platform for autonomous AI agents.

한국어 | 简体中文 | 日本語 | Français | Español | Deutsch | Português

License: MIT
Built with Tauri
Rust

LibrAgent is a local-first agent runner designed to maintain context across interactions. Unlike stateless clients, it keeps browser tabs and terminal sessions alive between turns, allowing agents to work more fluidly within a persistent workspace.

It implements open standards like MCP (Model Context Protocol) and Skills to remain modular and extensible.

Why LibrAgent?

The goal of this project is to make autonomous agents accessible. Many existing tools remain trapped behind terminal commands and manual JSON configurations, creating a gap that excludes many potential users. LibrAgent aims to bridge this gap by providing a local-first environment where anyone can deploy and manage agents without needing to be a developer.

🎬 Demo

LibrAgent Demo

Browser automation and shell execution in a single, stateful workflow.

Core Features

1. Persistent Workspace

Agents operate within a long-lived environment rather than spawning fresh processes for every turn.

  • Live Webview: Real-time browser automation using Tauri webviews. Sessions and cookies persist across turns.
  • Unified Terminal: A persistent, sandboxed shell (Python/Node.js supported) that shares state with the workspace.

2. Multi-Agent Orchestration

LibrAgent allows agents to delegate tasks to specialized sub-agents.

  • Assistants: Manage agent profiles with unique system prompts and tool configurations.
  • Swarm Intelligence: Parent agents can spawn, message, and await results from sub-agents to solve complex tasks.

3. Extensibility

The platform is designed to be expanded via community standards.

  • Extensions (MCP): Full support for the Model Context Protocol. Connect to any MCP server instantly.
  • One-Click Presets: Curated catalog for GitHub, Brave Search, etc., available directly in the UI.
  • Skills & Playbooks: Reusable behavior snippets and structured workflow templates.

4. Autonomy & Scheduling

  • YOLO Mode: Optional autonomous execution for sensitive tools without manual approval.
  • Scheduled Tasks: Cron-based automation with workspace-specific targeting and automatic recovery.

5. Context & Metrics

  • @mentions: Direct injection of files, skills, or playbooks into chat.
  • Multimodal: Handles images and audio for OpenAI, Anthropic, and Gemini models.
  • Observability: Real-time TPS metrics and prompt caching hits (for Anthropic/Gemini).

📖 Documentation & Guides

LibrAgent is a powerful platform, and we provide detailed documentation to help you get the most out of it.

  • Navigation Guide: A map of the application's structure, explaining how to use routes like /assistants (Assistant Profiles) and /playbooks (Workflow Templates) to manage your agents.
  • Architecture Guide: Detailed overview of how LibrAgent works under the hood.

📦 Installation

Download the latest binaries for Windows, macOS, or Linux from the Release page.

Build from source:

git clone https://github.com/fritzprix/libr-agent
cd libr-agent

For production desktop app builds (Tauri binaries/installers), resolve dependencies and run the Tauri build (which will run the frontend build via pnpm build):

pnpm install
pnpm tauri build

For local development, simply run the Tauri dev server. No need to run pnpm build before pnpm tauri dev — Tauri uses the Vite dev server (beforeDevCommand: pnpm dev), so an extra build just slows you down:

pnpm install
pnpm tauri dev

Design Choices

  • Local First: Your data and API keys stay on your machine.
  • Tauri + Rust: Chosen for security (memory safety), performance, and small binary size.
  • SQLite (SeaORM): Used for robust, local persistence of sessions and configurations.

Contributing & License

Contributions are welcome. Please see CONTRIBUTING.md.

License: MIT

Reviews (0)

No results found