📖About & Usage

About

Carol Scorer is a droid that watches user interactions across Carol’s surfaces and gives each one a score. Its job is to help Carol understand how well people are engaging — which conversations are helpful, which feel stalled, and where attention is needed. It turns raw chatter into a signal Carol can act on, without anyone having to read a hundred chat logs.

When it runs, Carol Scorer takes a user interaction — a WhatsApp message, a web chat reply, a command — and sends it to Claude for scoring. Claude reads the exchange and returns a numeric score and a brief reason, which Carol Scorer records for Carol to review later. The droid is a one-off, not part of any reusable family: it’s a custom tool built specifically for Carol’s needs, and no other agent runs a copy with the same name. It’s scheduled to run at regular intervals, though the exact timing is left at its owner’s discretion.

Because it calls Claude, Carol Scorer is not pure software — it leans on a language model to do the judging. The model it uses isn’t pinned down in the registry, so Carol may have configured it to use a default or a particular version. Either way, the droid’s real value is the scoring pipeline it automates, freeing Carol from manual monitoring and keeping her dashboard informed.

🛰️Updates

Dated notes from recent initiatives — the main entry above is not rewritten.

👤Owner

📚Recent initiatives

Initiatives that touched this droid — a short summary each; open one for the full story.