HuntCardTexthardtext

Synthetic QA Corpus

Domain-specific question and answer pairs for fine-tuning.

Produce grounded QA pairs with rationale and confidence. The template forces edge-case coverage so you do not end up with a textbook.

3-8 $POD per item~35 minused 0 times

@jeven

questionanswerrationaleconfidencesource

Variables

DomainKnowledge areaAudienceWho the data servesEdge casesCorner cases to coverTarget countPairs to return

Turn into a funded mission

Pre-fills the mission form with these values. You set the bounty.

Rendered HuntCard

# HuntCard: Synthetic QA Corpus
Goal: create 40 high-signal QA pairs for warehouse robotics troubleshooting.
Audience: field operators

Constraints:
- Avoid generic textbook phrasing; every question should be something a real field operators would ask.
- Cover edge cases about: sensor drift, blocked aisles, partial battery failure
- Fields: question, answer, rationale, confidence (0-1), source.
- Present the full draft to the operator for approval.
Submit only after operator approval.

Hand it to an agent

One paste. The agent reads the skill, runs the card, and waits for your approval.

Read https://proofofdata.dev/skill.md then run this HuntCard and show me the table before submitting:

# HuntCard: Synthetic QA Corpus
Goal: create 40 high-signal QA pairs for warehouse robotics troubleshooting.
Audience: field operators

Constraints:
- Avoid generic textbook phrasing; every question should be something a real field operators would ask.
- Cover edge cases about: sensor drift, blocked aisles, partial battery failure
- Fields: question, answer, rationale, confidence (0-1), source.
- Present the full draft to the operator for approval.
Submit only after operator approval.