
Egent
@xuezhaojun
A collaborative MCP server for sharing AI agent contexts between engineers
Overview
What is Egent?
Egent is a collaborative MCP server that converts a GitHub repository of AI agent contexts into an MCP Server. It enables engineers using different programming tools (Cursor, Windsurf, Cline, etc.) to share, version, and collaboratively improve task templates and agent instructions.
How to use Egent?
Configure Egent as an MCP server in your tool using npx -y egent@latest with either --context-repo for a remote GitHub repository or --context-path for a local directory. After configuration, issue the egent_start command to initiate interaction with your code-agent.
Key features of Egent
- Converts a GitHub repo into an MCP Server
- Collaborative editing through standard GitHub workflows
- Version control for AI prompts and task templates
- Immediate sharing across the entire team
- Supports both remote and local context sources
Use cases of Egent
- Build a task library for common operations like setting up new components
- Share team-specific knowledge and project patterns as agent contexts
- Implement authentication flows with middleware and API endpoints
- Refactor code across multiple files following consistent patterns
FAQ from Egent
What is the first command to use with Egent?
After configuring Egent, you must issue egent_start to initiate interaction with your code-agent.
Can I use a local directory instead of a remote GitHub repo?
Yes, use the --context-path argument with a local directory for testing or development.
What problem does Egent solve?
Egent solves context sharing across different programming tools, preventing knowledge silos and enabling version-controlled collaborative improvement of agent prompts.
What dependencies does Egent require?
Egent runs via npx, so Node.js and npm are required. No additional runtime or database is mentioned.