new·The score now tells you which way it movedA brain's exam only ever grows: its own material writes questions, and so does every question a real caller asked and did not get answered. The score is a percentage over that growing set, so a brain that learned more could post a smaller number — and this week three did. One of them answered two MORE questions than the week before and showed eighteen points less. Printed as a single percentage, that reads as decline to a reader and as punishment to anyone who contributes material.all news →
mozg.beta
Sign in

Building With AI Agents · all subjects

sources/crawl

3 notes, read out of this brain and free to use. Each one was extracted from a source and is re-checked against its exam.

Crawl one site: downloads from GitHub tree, llms.txt, sitemap, link traversal; only newest version

The site crawler (lib/crawl.ts) accepts one domain, downloads docs from: GitHub repo tree (if found), llms.txt, sitemap.xml, or by traversing links. Returns only the newest version. If it finds a 'versionscript' shell in the HTML, follows the link to the repo.

Site source expands to URL sources, avoids duplicates

When a user adds a site-type source (e.g., docs.example.com), the crawler (worker/crawl.ts) expands it into individual url-type sources. Repeating the same site crawl doesn't duplicate URLs—urls are de-duplicated.

Site without goal generates goal from material

If a site is added to a brain without a goal, the crawler extracts one from the documentation's title and README. Otherwise the goal stays blank until the user sets it.

Give your agent this brain