Synapse v0.1

Synapse is a general-purpose autonomous coding agent with a terminal interface. It can help build, debug, refactor, test, review, secure, and analyze software projects across multiple programming languages.

The agent uses Deep Agents and LangChain with an NVIDIA-compatible OpenAI API endpoint. It stores generated project files in a local workspace directory.

Requirements

• Python 3.10 or newer
• An NVIDIA API key with access to the configured model
• Windows, macOS, or Linux

Installation

Create and activate a virtual environment:

python -m venv .venv

Windows PowerShell:

.\.venv\Scripts\Activate.ps1

macOS or Linux:

source .venv/bin/activate

Install the dependencies:

pip install -r requirements.txt

Configuration

Create a .env file in the project directory:

NEMOTRON_API_KEY=your_api_key_here

Optional settings:

NVIDIA_BASE_URL=https://integrate.api.nvidia.com/v1
MODEL_NAME=nvidia/nemotron-3-super-120b-a12b
WORKSPACE_DIR=./workspace

NEMOTRON_API_KEY is required. Do not commit .env or expose the API key in source control.

Running Synapse

Start the interactive CLI:

python synapse_v0.1.py

Then enter a task in natural language, for example:

Build a FastAPI backend with authentication.

Review this repository for security vulnerabilities.

Refactor this TypeScript application and add tests.

CLI Commands

Command	Description
help	Show available commands and examples
workspace	Display the active workspace directory
clear	Clear the terminal
quit, exit, q	Exit Synapse

How It Works

• The main agent receives tasks through the terminal.
• The name_project tool creates a filesystem-safe project directory under the configured workspace.
• Specialized subagents handle code review, security review, testing, debugging, and architecture review.
• Conversation state is kept in memory for each task using LangGraph checkpointing.
• Agent-generated files are restricted to the configured workspace backend.

Project Structure

.
├── synapse_v0.1.py   # Main CLI and agent configuration
├── requirements.txt  # Python dependencies
├── .env              # Local API configuration; keep private
└── workspace/        # Generated projects, created automatically

Notes

• The default model and API URL can be changed through environment variables.
• The workspace directory is created automatically when the program starts.
• If a task fails, Synapse does not automatically roll back files. Inspect the workspace before continuing.