How Safe Source Discovery Should Work

A loop directory should not publish scraped text directly. Public source discovery is useful, but every candidate should go through filtering, scoring, classification, and human review before it becomes a public page.

A safe pipeline looks like this:

1. Search public code sources server-side.
2. Fetch candidate files within rate limits.
3. Hash content to deduplicate.
4. Score for goal, checks, exit conditions, and guardrails.
5. Use an optional AI classifier to extract structure and risks.
6. Keep the candidate in an admin queue.
7. Publish only after human review and editorial improvement.

The public page should add value: explanation, safety notes, attribution, structured steps, and copy-ready Markdown. That is what separates a useful directory from copied clutter.