Skip to content

Improve skill discovery for agents#138

Merged
marc0olo merged 1 commit intomainfrom
improve-skill-discovery
Mar 31, 2026
Merged

Improve skill discovery for agents#138
marc0olo merged 1 commit intomainfrom
improve-skill-discovery

Conversation

@marc0olo
Copy link
Copy Markdown
Member

Summary

Agents were failing to fetch skills because they constructed wrong URLs from the {name}/SKILL.md template in llms.txt — inventing filenames like internet-identity.md or index.json instead. The IC asset canister's SPA fallback returns HTML with 200 OK for these wrong URLs, giving agents no clear error signal.

  • Add url field to each skill entry in index.json so agents follow direct links instead of constructing URLs
  • Improve llms.txt with numbered steps, concrete example, and guardrails ("do not construct URLs manually", "if you get HTML, the URL is wrong")
  • Skip dotfiles/dirs in collectFiles to exclude .score_cache from the index

See #137 for the planned v0.2 spec migration (path rename to /.well-known/agent-skills/, schema changes).

🤖 Generated with Claude Code

- Add url field to each skill entry in index.json so agents follow
  direct links instead of constructing URLs from templates
- Improve llms.txt with numbered steps, concrete example, and
  guardrails against wrong URL construction
- Skip dotfiles/dirs in collectFiles to exclude .score_cache from index
@marc0olo marc0olo requested review from a team and JoshDFN as code owners March 31, 2026 10:53
@github-actions
Copy link
Copy Markdown

Skill Validation Report

No skill files were changed in this PR — validation skipped.

@marc0olo marc0olo merged commit 188f303 into main Mar 31, 2026
6 checks passed
@marc0olo marc0olo deleted the improve-skill-discovery branch March 31, 2026 11:31
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants