Synapticity is an autonomous development platform that manages the entire software lifecycle. It acts as a digital engineering team, coordinating specialized agents to handle planning, coding, security auditing, and deployment. By combining local processing (via Ollama) with cloud-based fallback (via Gemini), it provides a secure and resilient environment for software production.
Synapticity is more than just a code generator; it implements a structured workflow modeled after professional engineering standards. For the best experience, keep these principles in mind:
While the system is autonomous, it is designed with a "Human-In-The-Loop" approach.
- Review Before Approving: The system pauses at major milestones (like architectural planning and implementation). Always check the
MISSION_LOG.mdin your workspace before authorizing the next phase. - Audit the Path: Use the
./sync auditcommand to see exactly how the agents arrived at a decision.
Autonomous agents perform best when their goals are clear and well-defined.
- Start Small: Begin by generating foundational scaffolding.
- Refine as You Go: If the project needs to pivot, use the
./sync updatecommand to re-draft specifications without losing your progress.
You don't always need to run a full development cycle. You can dispatch a single agent for specific tasks, like refactoring a file or expanding documentation, using the standalone agent command.
Synapticity is optimized for users who want to bypass private API costs associated with tools like Claude or GPT.
- Google's Gemma (or Qwen/Llama): Acts as the open-weights "engine," which is free to download and run.
- Ollama: Used to run these high-IQ models locally on your machine with zero latency.
- The Framework Advantage: Synapticity is natively configured to use Ollama's local endpoints as the primary driver, providing a "Claude Code"-like experience without the Anthropic subscription fees.
Install the core dependencies:
pip install google-genai requests python-dotenv rich pydantic-settings cryptography bandit pytest- Local LLM: Install Ollama and ensure the service is running locally (
http://localhost:11434). - Environment: Copy
.env.exampleto.env. - Cloud Fallback: Add your Gemini API keys if you want a fallback for complex reasoning tasks.
Run the doctor command to verify everything is connected correctly:
./sync doctor| Command | Usage Example | Description |
|---|---|---|
| dash | ./sync dash |
View active missions and their current status. |
| launch | ./sync launch auth-svc "Build a JWT API" |
Start a new development cycle. |
| resume | ./sync resume auth-svc |
Continue a paused mission. |
| update | ./sync update auth-svc "Add OAuth2 support" |
Adjust the objective and re-start the planning phase. |
| audit | ./sync audit auth-svc |
Review the internal task list and upcoming queues. |
| agent | ./sync agent auth-svc swe "Refactor utils.py" |
Send a specific agent for a one-off task. |
| remove | ./sync remove auth-svc |
Delete a mission workspace. |
| review | ./sync review bug-fix "security audit" C:/my-project |
Audit and patch an existing local project. |
| learn | ./sync learn auth-svc |
Analyze mission history to improve future agent performance. |
| deploy | ./sync deploy auth-svc |
Create a GitHub repo and push the verified codebase. |
Synapticity is built on a "Zero-Defect" philosophy:
- Sandboxed Execution: All generated code is tested in an isolated environment before it ever touches your production workspace.
- Security Scans: Passively runs vulnerability analysis (Bandit) against all implementation blocks.
- Self-Healing: If code fails tests or scans, specialized QA agents identify the issue and instruct the Software Engineer to apply a fix automatically.
For more technical details, refer to the following resources in the technical_documents directory:
- Architecture Map: System design and state machine.
- White Paper: The philosophy of the framework.
- Free Setup Guide: Running things for zero cost.
- Project Roadmap: Upcoming features and known issues.