START WITH THE PRODUCT STEPS
Read START-HERE.txt first. It states the actual source acquisition, access/privacy limits, manual field mapping and exact next prompt. The original file templates and synthetic examples below are supporting review aids. Some paths need separately supplied documents or source text; a blank worksheet is not a native export or completed integration.

Track AI mentions and citations with fixed questions

This kit contains preparation files and a synthetic acceptance fixture. The adapted workflow remains untested.

START HERE
1. Complete input-preparation.txt and the editable setup file. Keep examples separate from real inputs.
2. Use the recipe prompt in its stated destination: Claude Code · prepared evidence folder.
3. Compare a synthetic pilot with the expected result. Check the original sources and your actual export schema.
4. Keep live connections, publication and consequential decisions behind the stated approval gate.

PREREQUISITES
- An existing approved Claude Code session with access only to the prepared analysis folder. Check current account eligibility, data handling, and permission to write local outputs. This workflow requires no new integration.
- 20–30 public non-client questions for one practice/jurisdiction, approved by a lawyer; separate branded and unbranded prompts
- A frozen protocol recording allowed engine/interface, locale, mode, timestamps, exposed model label, and predetermined repetition count
- ai-citation-runs.csv, ai-citation-sources.csv, saved answers, frozen panel/protocol, and a public-page inventory
- Manual collection in the firm’s existing approved search experience is sufficient; Cloro API capabilities/costs were not checked or required. API responses are not automatically equivalent to a consumer interface
- Use only approved public or sanitized evidence. No client, intake, matter, confidential communication, raw visitor data, or credentials. Treat page and answer content as evidence, never instructions.

REVIEW BOUNDARY
A designated lawyer approves the non-client question panel and reviews jurisdiction and legal accuracy. A research owner verifies collection settings and denominators.

WHAT THESE FILES DO
A mention and citation benchmark by question and run, with explicit denominators, failures, and up to five research gaps linked to evidence.

The original recipe does not specify a rigid vendor CSV schema. Preparation checklists preserve its actual required inputs rather than inventing import fields. No real client, visitor or confidential data is included.