ishandutta2007/Velocity

What it solves

Open Velocity is an open-source, web-native IDE designed to prevent developer lock-in to proprietary AI coding platforms. It provides a universal gateway that allows developers to use any LLM (such as GPT-4, Claude 3.5, Gemini 1.5 Pro, and Llama 3) to power autonomous AI agents that can plan, write, and verify code.

How it works

The platform uses a modular, container-native architecture consisting of a Web UI, a Gateway and Orchestrator, a Workspace Manager that handles isolated Docker containers, and an AI Model Gateway. Its backend core (OpenCentravity) manages concurrent multi-agent coordination through:

• Hierarchical Swarms: Tracking parent-child relationships between agents for complex workflows.
• Whiteboard Message Bus: A shared communication system for sub-agents to coordinate efforts.
• Distributed Lock Manager: File-level mutexes to prevent parallel agents from colliding when editing the same files.
• Memory Snapshots: A system to save agent states, allowing them to restore and self-heal from failures.
• Cost Tracking: Real-time monitoring of token counts and API costs per LLM call with configurable spending caps.

Who it’s for

Developers who want a model-agnostic, self-hosted AI coding environment where they can delegate complex tasks to autonomous agents without being tied to a single AI provider.

Highlights

• Universal LLM Gateway: Connects to a wide variety of models including Deepseek and Gemini.
• Multi-Agent Collaboration: Ability to spawn multiple agents working in parallel on different project parts.
• Verifiable Artifacts: Agents produce task lists, screenshots, and test results to verify their work.
• Enterprise-grade Backend: Includes a robust SQLite database layer with sequential migrations and connection safety mechanisms.