Turn your local LLM into an overqualified file clerk

What it does

ai-renamer is a Node.js CLI that asks a local or remote LLM to suggest sensible names for your files based on what they contain. It extracts frames from videos via ffmpeg, sends images to vision models like Llava, and renames other files by their contents. Your preferences—provider, model, case style, and character limit—are automatically saved to ~/ai-renamer.json.

The interesting bit

This is fundamentally glue code—ffmpeg, an LLM API, and the change-case library wired together—but the wiring is practical. By pulling video frames and routing images to vision-capable models, it turns a generic chat interface into an offline-capable batch renaming utility.

Key highlights

• Supports local inference via Ollama or LM Studio, plus OpenAI for cloud-based renaming
• Handles videos (via ffmpeg frame extraction), images, and general files
• Enforces consistent naming conventions with change-case (kebab, camel, snake, etc.)
• Remembers flags like model choice, max filename length, and output language in ~/ai-renamer.json
• Offers a custom prompt flag to nudge the model toward specific descriptions

Caveats

• Requires separately installed dependencies: an LLM server (Ollama or LM Studio) and ffmpeg for video support
• First-run auto-selection of the Llava model may fail, forcing manual specification

Verdict

Useful if your Downloads folder is a graveyard of IMG_0001 and screen-recording-final-final. Skip it if you want a fully self-contained binary or already have a rigorous naming scheme.