binquery: a local CLI shortlist, now also an agent skill

A developer has released binquery, a local Python CLI that indexes a user's video folder with ffmpeg and OpenCLIP, then encodes a natural-language query to return a shortlist of 8–15 candidate clips with path, score, gate, and reasons. The tool is designed to avoid full-library skims or auto-edits, and it can also be installed as an agent skill via npx or by copying the skill folder into .agents/skills or .claude/skills.

I cut from a local footage box. I wanted 8–15 candidates from one sentence, not a full-library skim and not an auto-edit. binquery is a local Python CLI. You index your own folder with ffmpeg + OpenCLIP. query encodes only the sentence and prints a shortlist: path, score, gate, reasons. Limit is 8–15. People still watch. It is not a browser. It is not on PyPI. Clone, then: python3 -m venv .venv .venv/bin/pip install -e . Needs ffmpeg / ffprobe on the machine. Agents can install the instructions as a skill this does not install the Python package : npx -y skills add jennifferweslowski-design/binquery Or copy skills/binquery/ into .agents/skills/binquery or .claude/skills/binquery .