Ellama

Ellama is an AI assistant for Emacs users who want help with writing and code inside their editor. It handles requests about selected text as well as longer conversations and coding tasks that need several steps.

• Text and code editing: Translate, summarize and proofread text, generate or revise code, review a buffer or selection, and draft commit messages from diffs.
• Project context: Include buffers, files, directories and Emacs Info nodes in a request, plus images and audio when the selected provider supports them.
• Persistent conversations: Save and resume chat sessions, switch between them, and compact older context manually or automatically as conversations approach the model's context limit.
• Coding agents: Agents can plan tasks, call tools and delegate work to subagents, using project instructions from AGENTS.md and reusable skills or prompt templates.
• Tool safeguards: Project policies, filesystem checks, edit hooks and data loss prevention scans can warn about or block reads, edits, shell commands and tool output.

Ellama runs inside Emacs and is free, open-source software under GPL-3.0. It connects to local models through Ollama, GPT4All and Llama.cpp, or to cloud services including OpenAI, Claude, Gemini and Vertex AI. It also supports OpenAI-compatible APIs. Local backends run model inference on your hardware; cloud providers receive the requests and context you send. Image input, audio input and tool calling depend on the selected provider's capabilities.