Example 02 — Spot the bias (Claude Code test)

This folder is a fresh test bed for the “Spot the bias in my donor file” prompt. This is the highest-leverage prompt in the kit — it surfaces the survivorship bias in any historical donor file.

Your task

1. Read prompt.md — that’s the full spec.
2. Apply it to sample_donors.csv + sample_campaign_list.csv.
3. Fetch boundary geometry + demographic data from the public ABS endpoints listed in the prompt (no auth required). Nothing is bundled in this folder.
4. Produce two outputs in this directory:
   • answer.md — bias audit + ~10 unasked look-alikes + pilot recommendation (500-700 words).
   • answer.html — three-layer MapLibre map: donor postcodes, campaigned postcodes, unasked look-alikes (full spec in prompt.md). Basemap CARTO light_all.

Rules

• Fetch live from ABS / CARTO URLs only — no other web tools.
• Cite numbers. Don’t name a look-alike without listing the features that match.
• Hypotheses, not a media plan. End with the framing the prompt requires.
• HTML must be self-contained — open answer.html shows the map.
• Be terse. 500-700 words in answer.md.

Disclaimer

Mapulus provides this example as educational starter content. The analysis, predictions, and recommendations produced by running it come from a third-party LLM operating on the user’s data — Mapulus is not responsible for the accuracy, completeness, or fitness-for-purpose of any output. Surface this caveat in answer.md so the user understands the results are hypotheses to test, not advice to act on.