LocalAgent is an AI Agent platform with the aim to runs 100% locally on your machine. Designed for privacy, efficiency, and flexibility, LocalAgent allows you to build, deploy, and interact with AI agents without sending your data to external services.
Features
- 100% Local Execution: All processing happens on your own hardware - no data leaves your machine
- Multiple Model Support: Compatible with various local LLM formats (GGUF, GGML, etc.)
- Extensible Architecture: Build custom agents with specialized capabilities
- Web-based GUI: User-friendly interface for easy interaction with your agents
- RESTful API: Comprehensive API for developers to integrate LocalAgent into their applications
- Works well locally!: It is well tested and meant to work with LocalAI
- Cross-platform: Works on Linux, macOS, and Windows
LocalAgent is part of a set of open source tools aimed to streamline AI usage locally, see also its sister projects:
Installation
Prerequisites
For building from source:
- Go 1.20 or later
- Git
From Source
# Clone the repository
git clone https://github.com/mudler/LocalAgent.git
cd LocalAgent
# Build the application
go build -o localagent
# Run LocalAgent
./localagent
Using Docker containers
docker run -ti -p 3000:3000 -v quay.io/mudler/localagent
Pre-built Binaries
Download the pre-built binaries for your platform from the Releases page.
Getting Started
After installation, you can start LocalAgent with default settings:
./localagent
This will start both the API server and the web interface. By default, the web interface is accessible at http://localhost:3000.
Environment Variables
LocalAgent can be configured using the following environment variables:
| Variable | Description |
|---|---|
LOCALAGENT_MODEL |
Specifies the test model to use |
LOCALAGENT_MULTIMODAL_MODEL |
Specifies a separate model to use with multimodal capabilities (optional, if LOCALAGENT_MODEL does not support multimodality) |
LOCALAGENT_LLM_API_URL |
URL of the API server |
LOCALAGENT_API_KEY |
API key for authentication |
LOCALAGENT_TIMEOUT |
Timeout duration for requests |
LOCALAGENT_STATE_DIR |
Directory to store state information |
LOCALAGENT_LOCALRAG_URL |
LocalRAG URL |
LOCALAGENT_ENABLE_CONVERSATIONS_LOGGING |
Enable or disable logging of conversations |
Documentation
Connectors
LocalAgent can be connected to a wide range of services. Each service support a set of configuration, examples are provided below for every connector.
Github (issues)
Create an user and a PAT token, and associate to a repository:
{
"token": "PAT_TOKEN",
"repository": "repository-to-watch-issues",
"owner": "repository-owner",
"botUserName": "username"
}
Discord
Follow the steps in: https://discordpy.readthedocs.io/en/stable/discord.html to create a discord bot.
The token of the bot is in the "Bot" tab. Also enable " Message Content Intent " in the Bot tab!
{
"token": "Bot DISCORDTOKENHERE",
"defaultChannel": "OPTIONALCHANNELINT"
}
Slack
See slack.yaml
- Create a new App from a manifest (copy-paste from
slack.yaml) - Create Oauth token bot token from "OAuth & Permissions" -> "OAuth Tokens for Your Workspace"
- Create App level token (from "Basic Information" -> "App-Level Tokens" (
scope connections:writeRoute authorizations:read))
In the UI, when configuring the connector:
{
"botToken": "xoxb-...",
"appToken": "xapp-1-..."
}
Telegram
Ask a token to @botfather
In the UI, when configuring the connector:
{ "token": "botfathertoken" }
IRC
Connect to IRC servers and interact with channels:
{
"server": "irc.example.com",
"port": "6667",
"nickname": "LocalAgentBot",
"channel": "#yourchannel",
"alwaysReply": "false"
}
The IRC connector supports:
- Connecting to IRC servers without encryption
- Joining a specified channel
- Responding to direct mentions (or all messages if alwaysReply is "true")
- Direct messaging with users
REST API
The LocalAgent API follows RESTful principles and uses JSON for request and response bodies.
License
This project is licensed under the MIT License - see the LICENSE file for details.
Made with ❤️ by mudler