`categories.py` runs them all and classifies their episodes into events, one line per event; `data.py` reads a station folder (`displacement/`, `metrics/`, `camera/`).
The repository is also a Claude Code plugin (`.claude-plugin/plugin.json`). Its workflow improves the detectors on a range of data where they perform poorly, given labels as the truth on that range:
1. measures before: the score on the range, and a benchmark on a reference truth (every labelled station);
2. the `upgrader` agent compares the detectors' events with the labels and fixes the detectors with general rules, one commit per fix;
3. the `validator` agent, an LLM-as-judge that does not read the upgrader's diagnosis, keeps the fixes, removes some, or cancels them: no station may go back, no overfitting;
The caller gives the context of each run: the range, the labels, the reference truth, the families document, any context documents, the output folder, and its own rules (git branch, what else to commit). For example:
Labels: JSONL, one per line, `{"id", "debut", "fin", "categorie", "description"}` (UTC ISO times, `categorie` = the family).
-`skills/wf-improve/SKILL.md`: the workflow; its tools in `skills/wf-improve/scripts/`: `score.py` (the score of a range), `regression.py` (the benchmark), `sensors.py` (the sensors of each event).
Copied from Support-Agent on 2026-11-13 (commit 48a3092); the detectors now evolve here. Used by Diagnostic-app and Optimizer, where it is the `detectors/` submodule; their labels come from the `labeling` repository, their data from `aimsight-data`.