A Claude Code plugin for LLM-driven text classification: apply a category set to a corpus via Claude or GPT, with prompt caching, structured outputs, and async or batch execution. Optionally hand-label a sample first and tune the prompt against it.
The plugin is the installable artifact — see plugin/README.md for the full documentation, install instructions, and usage.
Through Claude Code's marketplace mechanism. From a directory you trust:
/plugin marketplace add emilysilcock/econ-nlp-plugins
/plugin install text-classification@econ-nlp-plugins
After install, the slash commands /classify-run, /classify-tune, /classify-label, and /classify-report-issue become available (possibly namespaced as /text-classification:classify-run, etc.).
This plugin is a hard dependency of agentic-clustering — installing agentic-clustering auto-installs text-classification too. After /cluster-finalize produces a categories.json, the classify skills pick it up automatically. You can also install text-classification on its own for any classification task.
MIT — see LICENSE.