fix(gaia-skill): drop nonstandard version frontmatter key

[lsetiawan]

Summary

• The agentskills.io skill schema allows only specific frontmatter keys; consumers that enforce it — including inspect_ai.tool.read_skills, which the upcoming eval harness uses — reject download-script-dev/SKILL.md with Additional properties are not allowed ('version' was unexpected).
• Removes the version: 2026-03-20 key from that skill's frontmatter. The value was a redundant date stamp: the plugin's canonical version lives in plugin.json (0.1.0), and edit history lives in git, so nothing is relocated.
• Audited all 76 SKILL.md files under plugins/ and community-plugins/: this was the only file with a nonstandard key (every other file uses only name, description, and metadata). After the fix, all 76 validate cleanly.

Changes

• community-plugins/gaia-data-downloader/skills/download-script-dev/SKILL.md: delete the version: frontmatter key (1 line).

Test plan

• read_skills on every skill directory under plugins/ and community-plugins/ reports 76/76 valid (was 75/76, with this file the sole failure). Reproduce with:

  uv run --with inspect-ai python -c "
  from pathlib import Path
  from inspect_ai.tool import read_skills
  dirs = [p.parent for b in ('plugins','community-plugins') for p in Path(b).rglob('SKILL.md')]
  print(len(read_skills(dirs)), 'valid of', len(dirs))"

• CI validate-plugins workflow passes.

🤖 Generated with Claude Code

[github-actions]

🔍 Tessl Skill Review

community-plugins/gaia-data-downloader/skills/download-script-dev/SKILL.md

⚠️ Error: ✘ Skill review requires you to be logged in. Run tessl login to log in, then try again.

Note: tessl skill review is deprecated – use tessl review run instead.
Migration guide: https://docs\.tessl\.io/improving\-your\-skills/migrate\-from\-skill\-review

---

To improve your score, point your agent at the Tessl optimization guide. Need help? Jump on our Discord.

Feedback

Report issues with this review at tesslio/skill-review, or send private feedback from your terminal with tessl feedback.