Eigent: Multi-Agent Workflow Desktop App with CAMEL and MCP

Eigent is a desktop application designed for executing multi-agent workflows to automate complex, multi-step tasks. Built on top of the CAMEL-AI framework, it is designed to run locally, ensuring your data remains on your own machine.

The application distributes work across specialized agents tailored for specific tasks. Developer agents write and debug code, search agents scan the web for information, document agents synthesize reports, and multimodal agents process images and audio. These agents collaborate in parallel to accelerate project completion.

Eigent provides native support for Model Context Protocol (MCP) tools. This allows agents to browse the web, execute code, and connect to platforms like Notion, Google Workspace, or Slack. Users can also add custom integrations to equip their agents with the exact functional capabilities required for a specific job.

To ensure reliability, a human-in-the-loop mechanism intervenes if a workflow reaches an impasse. When the system encounters an ambiguity or a blocker, it prompts the user for input to guide the agent back on track.

Get Started

Cloud Version

The most efficient way to experience Eigent is through the cloud-hosted platform. It manages the models, APIs, and storage for you, allowing you to build and deploy workflows in minutes.

• Instant access to all features.
• Fully managed infrastructure.
• Premium support for subscribers.

Visit eigent.ai to claim free credits and start exploring.

Self-Hosted (Community Edition)

The Community Edition is ideal for teams requiring total local control and strict data privacy.

• Sensitive information never leaves your internal infrastructure.
• Freedom to modify and extend the platform to suit your specific needs.
• Eliminate recurring cloud subscription costs at scale.

Requirements: Node.js and npm.

Quick Start:

git clone https://github.com/eigent-ai/Eigent-desktop.git
cd Eigent-desktop
npm install
npm run dev

Enterprise Edition

The Enterprise Edition is designed for organizations with rigorous security, compliance, and customization demands.

• Dedicated commercial licensing.
• Exclusive features such as Single Sign-On (SSO) and bespoke development.
• Architected for scalable enterprise deployment.
• Negotiable SLAs and specialized implementation services.

Core Features

Workflows

In Eigent, a workflow is defined as a coordinated team of AI agents. The system deconstructs a broad objective and activates multiple agents to work on components of the task simultaneously.

Preset agent roles include:

• Developer Agent: Writes code and executes terminal commands.
• Search Agent: Navigates the web and extracts relevant content.
• Document Agent: Creates, organizes, and manages files and reports.
• Multimodal Agent: Analyzes and processes visual and auditory data.

Broad Model Support

When running Eigent locally, you have the flexibility to use your preferred Large Language Models (LLMs), whether they are proprietary or open-source.

MCP Tool Integration

A comprehensive collection of MCP-native tools is included with the application. These tools enable web browsing, code execution, and direct connectivity to Notion, Google Suite, and Slack. Users can also install custom tools or connect to internal APIs to further expand agent capabilities.

Human-in-the-Loop

If a workflow becomes stuck or requires a subjective decision, Eigent pauses execution and requests human intervention, ensuring the final output remains accurate and aligned with user intent.

100% Open Source

The source code for Eigent is entirely public. Users are encouraged to inspect, download, and modify the code. This community-driven approach ensures rapid innovation and transparency within the multi-agent ecosystem.

Use Cases

Palm Springs Tennis Trip Planner
Organize a 2026 trip for two tennis enthusiasts. The agents can research flights and hotels, create a detailed three-day itinerary—including activity times and costs—generate a full HTML report, and post the summary directly to Slack.

Q2 Financial Report from Bank CSV
The system can read a bank_transaction.csv file from your desktop and generate a professional HTML Q2 statement. This includes charts and data visualizations to show investors exactly how capital was allocated.

UK Medical Market Research Automation
Evaluate the feasibility of market entry for a premium electric scooter company in the UK. The workflow can size the market, map out local regulations, profile target consumers, and analyze distribution channels and pricing strategies to deliver a comprehensive HTML report.

German Electric Scooter Market Feasibility
Similar to the UK research, this workflow focuses on the specific regulatory and consumer landscape of the German market.

Multi-Agent Product Launch SEO Audit
Perform a comprehensive audit of a company website prior to a new workflow product launch. The agents provide a detailed optimization report featuring actionable fixes to improve search visibility.

Identify Duplicate Files in Downloads
Scan directories such as mydocs within the Documents folder to identify exact or near-duplicate files. The agent then groups them by similarity for easy management.

Tech Stack

Backend

• Framework: FastAPI
• Package Manager: uv
• Async Server: Uvicorn
• Authentication: OAuth 2.0, Passlib
• Multi-Agent Framework: CAMEL

Frontend

• Framework: React
• Desktop Wrapper: Electron
• Language: TypeScript
• UI Components: Tailwind CSS, Radix UI, Lucide React, Framer Motion
• State Management: Zustand
• Flow Editor: React Flow

Eigent delivers powerful multi-agent automation to your desktop. It is a local-first, open-source solution built for teams that prioritize control, privacy, and flexibility.