Colaberry·Library
⏳ 0 🔑 Sign in 📐
🌐

Ingest from URL or GitHub

Paste a webpage URL or a GitHub repo. The pipeline fetches → parses → classifies → enriches → submits.

🔗 Single URL

Webpage, raw markdown, JSON manifest, README, any single document.

🐙 GitHub repo (batch)

Walks the tree, processes 1 file at a time (up to 200), submits each to pending review.

🕘 Recent ingest batches

0 batches
No batches yet. Try ingesting a URL above.
🤖 What the pipeline does
  1. Fetch — HTTP GET (auto-rewrites github.com/.../blob/ → raw.githubusercontent)
  2. Parse — HTML title/meta/h1, Markdown frontmatter/h1/code blocks, JSON manifests
  3. Classify — routes to one of 15 categories with confidence + reason chain
  4. Enrich — extracts name, description, how-to-use, example, tags, version, owner
  5. Submit — lands in pending review queue (or auto-vets if source is in config/library_trusted_sources.json)

Per-item progress is observable on the batch progress page.