Jarvis Agent Platform - MVP

Personal AI agent platform built with FastAPI, MongoDB, Redis, and Celery.

Quick Start

Prerequisites

Docker & Docker Compose
Python 3.11+ (for local development)

Run with Docker Compose

# Copy environment template
cd docker
cp .env.example .env

# Start all services
docker-compose up -d

# Check health
curl http://localhost:8000/health

Local Development

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Start dependencies (MongoDB, Redis, Qdrant)
cd docker
docker-compose up -d mongodb redis qdrant

# Run API server
cd ..
uvicorn jarvis.app.main:app --reload

# Run Celery worker (separate terminal)
celery -A jarvis.app.tasks.celery_app worker --loglevel=info

Project Structure

jarvis/
  app/
    api/routes/        # FastAPI routers
    core/              # Settings, logging, app factory
    models/            # Pydantic DTOs
    services/          # Business logic facades
    runtime/           # Agent event loop & DAG executor
    tasks/             # Celery task definitions
    messaging/         # Redis Pub/Sub messaging
    llm/               # LLM gateway & tool registry
    integrations/      # Adapters & browser automation
    db/                # DB clients (Motor, Redis, Qdrant)
    utils/             # Helpers
  docker/              # Docker Compose & Dockerfile

API Endpoints

GET /health - Health check
POST /agents - Create agent
GET /agents/{agent_id} - Get agent state
POST /agents/{agent_id}/tasks - Submit task
GET /tasks/{task_id} - Get task status
POST /messages - Send message between agents

Documentation

CLAUDE.md - System Overview (Purpose, Architecture, Components, Flow)
IMPLEMENTATION.md - Implementation Progress & Status
SETUP_PROCEDURE.md - Development Phase Roadmap
HLD.txt - High-Level Design
LLD.md - Low-Level Design
Requirement.txt - Requirements

Environment Variables

See docker/.env.example for configuration options.

Key variables for LLM integration:

JARVIS_OPENAI_API_KEY – OpenAI key used by the agent runtime.
JARVIS_DEFAULT_LLM_MODEL – Override default model (defaults to gpt-4o-mini).

Development Status

Phase 3 Complete (75%) - Core agent runtime, tools, and MCP protocol implemented.

✅ Completed:

Agent runtime loop with Redis Pub/Sub messaging
LLM provider integrations (OpenAI, Gemini)
Tool calling system (5 tools: execute_code, calculator, get_time, web_search, read_url)
MCP protocol (server, client, tool discovery & execution)
Vector memory with Qdrant for long-term context
DAG executor for parallel task execution

⏳ Next Steps (Phase 4):

Agent discovery and registry
Inter-agent message protocol
Master-subagent collaboration patterns

License

Proprietary

twitch

Summary

Install to Claude Code