Gemini Srv

Gemini Srv is a web server and API that provides a user-friendly interface for interacting with the gemini-cli tool. It allows for persistent, context-aware conversations and includes a powerful scheduler for running autonomous, scheduled tasks.

Use Cases

• AI Assistant on a VM at your disposal: Deploy Gemini Srv on a virtual machine to have a dedicated AI assistant accessible from anywhere.
• Your custom Gemini-CLI agent integrated to Google Chat, Slack or Teams: Integrate the Gemini Srv API with your favorite communication platforms to bring your custom Gemini-CLI agent directly into your team's workflow.
• Automated Task Execution: Leverage the scheduler to run autonomous tasks that gather data, interact with the Gemini model, and perform actions based on predefined conditions.
• Scheduled Reporting and Alerts: Configure scheduled tasks to generate reports or send alerts based on specific criteria, powered by Gemini's analytical capabilities.
• Interactive Web UI for Gemini-CLI: Provides a user-friendly web interface to interact with your Gemini-CLI agent, manage conversations, and monitor scheduled tasks.

Features

• Web UI: A clean, modern web interface for managing conversations and tasks.
• REST API: A simple, stateless API for easy integration with other services like Slack, Teams, or custom scripts.
• Persistent Conversations: Conversation histories are saved to disk, so you can pick up where you left off.
• Scheduled Tasks: Define autonomous tasks in .toml files that can gather data and conditionally call the Gemini model.
• Context-Aware: Leverages gemini-cli's ability to read context from GEMINI.md files.

Getting Started

Prerequisites

• Go 1.18+
• A working gemini-cli installation.

Installation & Running

1. Clone the repository:

   git clone <repository_url>
   cd gemini-srv

2. Create a .env file: Copy the example file and customize it with your details.

   cp .env.example .env
   # Edit .env with your preferred user/pass and the correct path to gemini-cli

3. Build the server:

   go build .

4. Run the server directly (for testing):

   ./gemini-srv

5. Open your browser and navigate to http://localhost:7123.

Launching the A2A Server

Gemini-srv needs gemini-cli to be running the A2A server. You can get it running with these steps:

1. git clone https://github.com/google-gemini/gemini-cli.git && cd gemini-cli
2. npm install && npm run build
3. CODER_AGENT_PORT=41242 npm run start --workspace @google/gemini-cli-a2a-server

Running as a Service (systemd)

To run gemini-srv as a persistent background service that starts on boot, you can create a systemd unit file.

1. Create the service file: Create a new file at /etc/systemd/system/gemini-srv.service:

   sudo nano /etc/systemd/system/gemini-srv.service

2. Add the following content: Be sure to replace /path/to/your/gemini-srv with the actual absolute path to the project directory. The WorkingDirectory is crucial as it tells the service where to find the .env file, the static/ directory, and the data/ directory.

   [Unit]
   Description=Gemini Srv - A web UI and API for gemini-cli
   After=network.target
   
   [Service]
   Type=simple
   User=your_username # Replace with the user you want to run the service as
   Group=your_groupname # Replace with the group for that user
   
   # Set the working directory to the project root
   WorkingDirectory=/path/to/your/gemini-srv
   
   # Command to start the server
   ExecStart=/path/to/your/gemini-srv/gemini-srv
   
   Restart=on-failure
   RestartSec=5s
   
   [Install]
   WantedBy=multi-user.target

3. Enable and start the service:

   # Reload the systemd daemon to recognize the new file
   sudo systemctl daemon-reload
   
   # Start the service now
   sudo systemctl start gemini-srv
   
   # Enable the service to start automatically on boot
   sudo systemctl enable gemini-srv

4. Check the status and logs:

   # Check if the service is running
   sudo systemctl status gemini-srv
   
   # View the live logs
   sudo journalctl -u gemini-srv -f

API Usage

The server exposes a simple REST API for integrations.

• POST /api/v1/conversations: Create a new conversation.
• GET /api/v1/conversations: List all conversation IDs.
• GET /api/v1/conversations/{id}: Get the history of a conversation.
• POST /api/v1/conversations/{id}/prompt: Send a prompt to a conversation.
• DELETE /api/v1/conversations/{id}: Delete a conversation.

All API endpoints are protected by Basic Authentication using the credentials set in your .env file.