llmf
Installation
SKILL.md
llmf - Zero-Dependency Local LLM Runner
The llmf module leverages llamafile technology to run large language models locally with zero setup. It provides a full suite of tools for serving, chatting, and managing models.
When to Activate
- When the user wants to run an AI model locally without cloud API access.
- When setting up a local OpenAI-compatible API server (
serve). - When performing fast, one-off text generation tasks via CLI.
- When managing local GGUF or llamafile models (download, import, tokenize).
Core Principles & Rules
- Zero-Dependency: Emphasize that models run locally without external runtimes.
- Compatibility: The
servecommand provides an OpenAI-compatible HTTP interface. - Resource Management: Models are stored in
~/.x-cmd/data/llmf/model/.
Additional Scenarios
- Token Analysis: Use
tokenizeto breakdown text into token details. - Headless Server: Start the API server without opening a browser using
--nobrowser.