Swarm Orchestrator
An MCP server that extends AI coding assistants with deterministic, algorithmic capabilities such as code analysis, fault localization, and formal verification, enabling an autonomous engineering team within the IDE.
README
⚠️ DEPRECATED: Fully migrated to hermes agent.
<div align="center"> <img src="docs/images/swarm_logo.png" alt="Swarm Logo" width="200"/>
Swarm Orchestrator v3.4
Turn your IDE into an Autonomous Engineering Team
</div>
What is Swarm?
Swarm Orchestrator is a Model Context Protocol (MCP) server that extends AI coding assistants like Antigravity, Cursor, and Claude Desktop with deterministic, algorithmic capabilities. Instead of relying purely on LLM reasoning, Vexorbis Swarm uses specialized workers—code analysis, fault localization, formal verification, and autonomous Git management—to deliver faster, more reliable results.
Demo
https://vimeo.com/1158005784
✨ The Three Pillars
<table> <tr> <td width="33%" valign="top">
🧠 Algorithmic Core
Go beyond text search.
- HippoRAG: AST-based knowledge graphs with Personalized PageRank for deep code understanding.
- Ochiai SBFL: Find bugs statistically, not by guessing.
- Z3 Verifier: Formal verification for critical logic.
</td> <td width="33%" valign="top">
🤖 Autonomous Workforce
Let the agents handle the grunt work.
- Git Agent Roles: Specialized workers (Architect, Engineer, Auditor) that plan, commit, and create PRs.
- Dynamic Toolsmith: Vexorbis Swarm can design and register new tools at runtime.
- Markdown-Driven: Control the workforce from a
PLAN.mdfile.
</td> <td width="33%" valign="top">
🛡️ Active Governance
Stay in control.
- Telemetry Memory: SQLite-backed persistence for session context and failure analysis.
- Self-Healing: Detects repeated failures and alerts agents to change strategy.
- Permission-First: Autonomous actions (like tool creation) require user approval.
</td> </tr> </table>
🚀 Quick Start
Docker (Recommended)
git clone https://github.com/AgentAgony/swarm.git
cd swarm
docker compose up -d --build
Configure Your IDE
Add the MCP server to your IDE's configuration (e.g., ~/.gemini/antigravity/mcp_config.json on Windows):
{
"mcpServers": {
"swarm-orchestrator": {
"command": "docker",
"args": ["exec", "-i", "swarm-mcp-server", "python", "server.py"]
}
}
}
🔍 Tool Selection Guide
flowchart LR
Start([My Goal]) --> Q1{Type?}
Q1 -->|Find symbol by name| A["search_codebase(keyword_only=true)<br/>~1ms"]
Q1 -->|Find code by concept| B["search_codebase()<br/>~240ms"]
Q1 -->|Understand architecture| C["retrieve_context()<br/>~1-2s"]
Q1 -->|Debug failing tests| D["debug command<br/>Ochiai SBFL"]
style A fill:#2ecc71,color:#fff
style B fill:#3498db,color:#fff
style C fill:#9b59b6,color:#fff
style D fill:#e74c3c,color:#fff
🔄 Antigravity Sync
Vexorbis Swarm synchronizes with a human-readable PLAN.md file, enabling a true co-pilot workflow.
From Your Markdown:
## Todo
- [ ] Refactor auth module @architect
- Context: auth.py
- [ ] Add login tests @engineer
Vexorbis Swarm interprets:
[ ]→ Task isPENDING. Agent picks it up.@architect→ Assigns to the Architect worker.Context:→ Pre-loadsauth.pyinto the agent's window.
On completion:
- Agent marks the task
[x]in yourPLAN.mdautomatically.
📖 Documentation
| Getting Started | Concepts | Guides | Reference |
|---|---|---|---|
| Introduction | Architecture | PLAN.md Syntax | Tools |
| Installation | Decision Logic | Git Workflows | Configuration |
| Quick Start | Three Pillars | Debugging | API Reference |
| Custom Tools | Troubleshooting |
📊 Implementation Status
| Component | Status |
|---|---|
| HippoRAG (Multi-language AST) | ✅ Stable |
| Hybrid Search Engine | ✅ Stable |
| Ochiai SBFL | ✅ Stable |
| Telemetry & Memory | ✅ Stable |
| Git Agent Roles | 🚧 Experimental |
| Z3 Verifier | ⚠️ Partial |
| Structured Deliberation | 🚧 Stub |
🤝 Contributing
We welcome contributions! See CONTRIBUTING.md.
Development Mode:
# Enable debug features
export SWARM_DEBUG=true
export SWARM_VERBOSE_TELEMETRY=true
📜 License
MIT License - See LICENSE for details.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.