OmniContext
Universal State Manager & Handoff Bridge
Bridged the gap between AI coding assistants with a local daemon, CLI dashboard, and native MCP server for seamless session handoffs.
Context
OmniContext is a developer tool that bridges the gap between different AI coding assistants (Claude Code, Cursor, ChatGPT, Perplexity, etc.). By running locally as a daemon or a global CLI, it monitors the current project's Git state and active tasks, allowing developers to seamlessly handoff session context from one AI tool to another without losing context.
The problem
AI coding assistants operate in silos and often have usage limits. When a developer hits a limit in one tool (like Cursor) and needs to switch to another (like Claude Web), they lose all their context—active files, Git diffs, and current tasks. The challenge was building a universal bridge that captures this state and ports it seamlessly across different AI environments.
How I built it
Built a global CLI tool running locally via Node.js to monitor Git state and uncommitted changes
Created a Next.js dashboard that generates a highly-compressed Markdown 'Handoff Prompt' for web-based LLMs
Implemented a native Model Context Protocol (MCP) server for advanced IDE integration
Added tools for AI assistants to autonomously read the 'context://current' resource and manage tasks
Why these choices
MCP Server Integration
Implementing a native MCP server allows IDEs like Cursor and Claude Code to autonomously connect to the local daemon and read project context without manual copy-pasting.
Compressed Markdown Handoff
Web LLMs have context limits. Generating a highly-compressed Markdown representation of the Git state ensures the prompt fits within standard context windows while preserving crucial diffs.