feat(blueprints): create blueprints/users/Varunrnair/sakhi-non-expert-maternal-health-benchmark.yml

[Varunrnair]

Blueprint Contribution

Blueprint Details

• Blueprint ID: sakhi-non-expert-maternal-health-benchmark
• Category/Focus: public-health, maternal-health, healthcare-safety
• Models to test: CORE

What This Blueprint Tests

• Evaluates model understanding of maternal health topics using 231 community-sourced questions, with reference answers reviewed by healthcare nonprofit staff and ASHA workers, across English, Hindi, and Marathi (693 evaluation cases total).
• Tests whether responses align with deployment-oriented reference answers that reflect how maternal health questions are asked and answered in community health settings, rather than only textbook or clinical formulations.
• Assesses the model's ability to provide accurate, practical, and culturally appropriate guidance on pregnancy, maternal care, nutrition, infection prevention, and related health topics without promoting misinformation or unsafe advice.
• Measures consistency on public-health-relevant questions where incorrect information could contribute to real-world harm, using theme-specific rubrics comparable to the expert evaluation track.
• Evaluates multilingual performance on question distributions representative of field deployment contexts, including ASHA/home-visit concerns, healthcare access barriers, and local health practices.

Checklist

• My blueprint is in blueprints/users/<my-github-username>/ directory
• Blueprint YAML is valid and follows the [blueprint format](https://github.com/weval-org/configs/blob/main/README.md)
• Each prompt has a meaningful, descriptive id (e.g., france-capital-test, not p1 or auto-generated)
• Blueprint has clear success criteria (should assertions with specific criteria)
• I've used $not_* functions instead of should_not blocks where applicable
• I've tested the blueprint locally if possible (pnpm cli run <path-to-blueprint>)
• I agree to dedicate my contribution to the public domain under CC0 1.0 Universal

Notes

This blueprint expands CivicEval's public-health coverage by focusing on maternal health, a domain where inaccurate or misleading model outputs may have significant real-world consequences. The evaluation is intended to measure factual accuracy, responsible communication, and alignment with established maternal health guidance.

---

Automated Evaluation: This PR will trigger an automated evaluation with cost-controlled limits (max 10 prompts, CORE models only). Full evaluation runs automatically after merge.

• ✅ Validation: GitHub Actions will check YAML syntax and structure
• 🤖 Evaluation: Webhook will run limited evaluation and post results
• 📊 Results: View status and full analysis via links in comments

[claude]

Claude Code Review

This pull request is from a fork — automated review is disabled. A repository maintainer can comment @claude review to run a one-time review.

[weval-bot]

❌ Blueprint validation failed

• blueprints/users/Varunrnair/sakhi-non-expert-maternal-health-benchmark.yml: Failed to fetch blueprint content

[nojibe]

Hi @Varunrnair — same issue here as on #28: the automated evaluation failed with Failed to fetch blueprint content because this file is ~1.6 MB, and the GitHub Contents API doesn't return inline content for files over 1 MB, so our fetcher receives empty content.

Suggested fix: split the blueprint by language so each file stays well under 1 MB:

• blueprints/users/Varunrnair/sakhi-non-expert-maternal-health-en.yml
• blueprints/users/Varunrnair/sakhi-non-expert-maternal-health-hi.yml
• blueprints/users/Varunrnair/sakhi-non-expert-maternal-health-mr.yml

Each file keeps its own config header (adjust title/description/tags per language). As a bonus, per-language results make the multilingual-parity comparison easier to read on the dashboard.

You can push the split files to this same branch and the PR will update in place. Thanks again! 🙏