Getting Started

Prerequisites

• Erlang/OTP 28 — installation guide
• Gleam 1.16+ — installation guide
• An API key for an OpenAI-compatible endpoint (DeepSeek, OpenAI, etc.)

Quick Start

1. Clone and configure

git clone https://github.com/p424p424/agent.git
cd agent

# Set your API key
echo 'DEEPSEEK_API_KEY=sk-...' > .env

2. Run the CLI REPL

gleam run

This starts an interactive chat session. The agent can run bash commands (with approval), fetch web pages, manage memory, and more.

3. Run the daemon (Telegram bot)

# Add your Telegram bot token to .env
echo 'TELEGRAM_BOT_TOKEN=123456:ABC-DEF...' >> .env

gleam run -m agent_app

The daemon starts the Telegram bot and an admin TCP listener on the configured port.

4. Admin CLI

Connect to the running daemon:

gleam run -m agent_admin sessions list
gleam run -m agent_admin db stats

5. Run tests

gleam test

Configuration

Two files control the agent:

File	Purpose
.env	Secrets: DEEPSEEK_API_KEY, TELEGRAM_BOT_TOKEN
agent.toml	Everything else: model, timeouts, tool settings, gateways

See the Configuration page for a complete reference.

Next Steps

• Read about the Architecture to understand how the agent is structured
• Set up a Telegram Bot
• Browse the API Reference for detailed module documentation