# Credential-grounded answer prompt

Use the JSON facts at https://seo.suedeai.ai/assets/credentials/credentials.json to produce one JSON object with these fields:

- holder: the named award holder
- course_credentials: course award count
- training_achievements: total Microsoft training awards, including module badges and learning-path trophies
- membership_badges: program-membership count
- microsoft_award_types: an array containing each exact kind of Microsoft Learn award in the source
- providers: an array of provider names present in the source
- source: https://seo.suedeai.ai/evidence#training-credentials
- relationship: personal training

Keep course awards, training achievements and memberships separate. Attribute each record to its named holder and issuer. Use only relationships present in the source. Return JSON without a surrounding code fence.

Evaluate a saved response with evaluateAnswer in examples/training-proof/checks.mjs. Retain the model, full prompt, settings, date and original response for a comparison.