Skip to content

Portable checklist

The portable checklist is for a human or an external browser agent collecting public evidence without the repository's scripts. It is generated from the canonical rubric.

Open the generated checklist:

It contains every check, its anchors, the evidence to capture, and the recommended intervention.

Rules for an external agent

  1. Record the audit mode: public.
  2. Do not use an authenticated Google session.
  3. Do not post, reply, book, submit a form, call, text, or change a profile.
  4. Cite evidence for every score.
  5. Use verified only for direct observation or safe measurement.
  6. Use strong_signal when the observation is real but the consequence is not proven.
  7. Owner-data questions never reduce the public score.
  8. Do not invent a field that the public profile does not show.
  9. Do not publish a real client audit without approval.

Minimum evidence set

  • Business name, place ID, trade, market, audit timestamp.
  • Profile URL and destination URL.
  • Primary and additional categories.
  • Phone and website actions.
  • Address or service-area model.
  • Regular hours.
  • Services and attributes, if publicly visible.
  • Rating, review count, and a recent review sample.
  • Media count and useful examples.
  • Action destinations and their resolution.
  • Source parameters, if visible.

Output

Return a completed audit.filled.json that validates against the schema, then run:

python3 .opencode/skills/home-service-gbp-audit/scripts/score.py --input audit.filled.json --out scores.json --strict

Do not hand-write scores.json.

Why a generated checklist

The checklist is tied to the rubric version. If the rubric changes, regenerate it:

python3 .opencode/skills/home-service-gbp-audit/scripts/render_portable.py

CI rejects a stale generated checklist.