efrit  by steveyegge

AI coding agent for autonomous Emacs development

Created 3 months ago
315 stars

Top 85.5% on SourcePulse

GitHubView on GitHub
Project Summary

Efrit is an AI-powered autonomous Emacs assistant that enables direct Elisp evaluation and agent-to-agent communication, positioning it as both a user assistant and a platform for AI-driven development. It allows AI agents to autonomously debug, test, and enhance Efrit's functionality, and supports multi-turn conversations and natural language command execution.

How It Works

Efrit leverages Emacs' native Elisp evaluation capabilities to execute AI-generated code and commands. Its core innovation is the agent-to-agent architecture, which uses a file-based queue system for communication between Efrit and other AI coding agents. This allows for autonomous development loops where AI agents can interact with, modify, and improve Efrit's own codebase. It also supports multi-modal interfaces including chat, command, and agent modes.

Quick Start & Requirements

  • Installation: Clone the repository and add the lisp directory to your Emacs load-path. Configure your Anthropic API key in ~/.authinfo.
  • Prerequisites: Emacs version 28.1 or later, Anthropic API Key, Internet Connection.
  • Setup: Restart Emacs and test with M-x efrit-chat.
  • Docs: QUEUE_USAGE_EXAMPLE.md, AUTONOMOUS_AI_PLAN.md, CONTRIBUTING.md.

Highlighted Details

  • Agent-to-Agent Architecture: Enables AI agents to communicate with Efrit via a file-based queue system for autonomous development.
  • Self-Enhancing Capabilities: AI agents can debug, test, and modify Efrit's source code.
  • Autonomous Development Mode: Supports long-running AI development sessions by spawning independent Emacs instances.
  • Direct Elisp Evaluation: Provides full access to Emacs' native programmability for AI agents.

Maintenance & Community

  • Development: The project is actively developing its autonomous AI development platform capabilities.
  • Contribution: Guidelines are available in CONTRIBUTING.md.

Licensing & Compatibility

  • License: Apache License, Version 2.0. This license is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

  • Efrit relies on the Anthropic API, specifically Claude models, for its AI processing. The project is in active development, with self-enhancement capabilities listed as the "Next Phase."
Health Check
Last Commit

1 month ago

Responsiveness

Inactive

Pull Requests (30d)
2
Issues (30d)
1
Star History
16 stars in the last 30 days

Explore Similar Projects

Starred by Andrew Ng Andrew Ng(Founder of DeepLearning.AI; Cofounder of Coursera; Professor at Stanford), Thomas Wolf Thomas Wolf(Cofounder of Hugging Face), and
4 more.

ag2 by ag2ai

1.1%
4k
AgentOS for building AI agents and facilitating multi-agent cooperation
Created 11 months ago
Updated 16 hours ago
Feedback? Help us improve.