Summary

Improve the GitHub landing-page presentation for HASH so a first-time reader can see representative handwriting scans immediately and understand the dataset's rights posture before digging into the JSONL indexes.

Changes

• Add CI, metadata-license, and scan-count badges at the top of the README.
• Add the required creator attribution near the top and the current Credits block at the bottom.
• Add a representative sample-scan grid under docs/assets/hash-sample-grid.jpg using records with public-domain / Public Domain Mark / Israel-public-domain rights in the scan index.
• Add an "At a Glance" table with current scan/source counts, corpus size, rights-policy link, and canonical index link.
• Keep the existing generated status block intact.

GitHub metadata

Updated directly on the repository card with gh repo edit:

• Description: Rights-clean dataset of scanned handwritten Hebrew-script documents for OCR/HTR research.
• Topics: hebrew, ocr, htr, handwriting, dataset, cc0

Validation

Passed:

• git diff --check -- README.md docs/assets/hash-sample-grid.jpg
• README/index/sample-grid smoke check: exactly two creator-credit occurrences, Credits at the bottom, 198 indexed entries, sample grid linked, canonical index linked, sample grid is 1260x840
• python3 scripts/validate_indexes.py: ok: 111 sources, 198 entries, 228 files verified, recipe ok

Full-suite note:

• Initial python3 -m pytest could not collect because pyarrow was missing locally.
• After python3 -m pip install -r requirements-dev.txt, python3 -m pytest ran and reported 74 passed / 6 failed.
• The failures are pre-existing data-artifact/release-fixture issues on this origin/main base, not introduced by this README/image-only PR:
  • stale exports/entries.csv
  • stale datapackage.json
  • test_builder_picks_original_among_multiple_roles
  • test_attribution_gate_is_license_driven

Base-history note

The local main checkout has four unpushed HTR/transcript commits. This branch was intentionally created from origin/main to keep this PR limited to presentation changes.