Discover and explore top open-source AI tools and projects—updated daily.
Code completion as-you-type for Emacs
Top 98.2% on SourcePulse
Minuet is an Emacs package that provides AI-powered code completion as-you-type, integrating with a wide range of Large Language Models (LLMs) including OpenAI, Gemini, Claude, Ollama, and Llama.cpp. It aims to enhance developer productivity by offering intelligent code suggestions directly within the Emacs environment, supporting both specialized chat-based prompts and Fill-in-the-Middle (FIM) completion for compatible models.
How It Works
Minuet leverages Emacs's extensibility to provide a seamless code completion experience. It supports multiple LLM providers through a unified interface, allowing users to configure their preferred model and API endpoints. The package offers two primary completion modes: a minibuffer frontend for traditional completion UIs and an overlay ghost text frontend for inline suggestions. It also features streaming support for faster feedback, especially with slower LLMs, and includes customizable options for context window size, request timeouts, and suggestion behavior.
Quick Start & Requirements
(package-install 'minuet)
.plz
0.9+, and dash
. An API key for at least one supported LLM provider is necessary.minuet-provider
and plist-put
with specific provider options (e.g., minuet-openai-compatible-options
). Keybindings are provided for invoking completions and managing suggestions. Auto-suggestion can be enabled via minuet-auto-suggestion-mode
.Highlighted Details
Maintenance & Community
Minuet is part of GNU ELPA, indicating a commitment to long-term maintenance. While substantial contributions require copyright assignment to the FSF, minor contributions are welcomed. Further details on contributions and acknowledgments of related projects are available.
Licensing & Compatibility
As part of GNU ELPA, Minuet is distributed under a free software license, likely GPLv3 or similar, which may have implications for use in closed-source projects due to copyleft provisions. Specific compatibility for commercial use would depend on the exact license and any associated API terms of service for the LLM providers used.
Limitations & Caveats
The README notes that Deepseek's high server demand may slow down default completions, recommending alternative providers. It also advises against using "thinking models" due to increased latency, and if used, their thinking capabilities should be disabled. Users experiencing setup issues are directed to check API key configurations or model/context window sizes that might cause timeouts.
3 days ago
1 day