Grok BotKnown issues

Grok Bot sees 0 stars and an empty README on GitHub pages

Staff reproduced empty star, fork, README, and file-list reads. Ask the Bot to use the GitHub API or its browser instead.

When Grok Bot opens a GitHub repo page, it can report "0 stars, 0 forks" and miss the README. On Oct 13 at 2:52 AM ET, Cursor staff member Colin said the team reproduced it: "the star and fork counts (and the README and file list) currently come through empty," which is why a bot called a popular repo untrustworthy. He flagged it to the team.

Why it matters#

Bots that judge a repo by stars, forks, or README contents can reject good libraries or invent risk from empty scrapes.

How to use it#

  1. Do not trust star or fork counts the Bot claims it read from a github.com HTML page right now.
  2. Ask it to call the GitHub API from its computer instead.
  3. Or ask it to open the repo in its own browser and read the counts from the live page.
Prompt
Check https://github.com/OWNER/REPO. Do not trust a plain page scrape for stars or forks. Run curl https://api.github.com/repos/OWNER/REPO on your computer and report stargazers_count and forks_count from the JSON. If that fails, open the repo in your browser and read the counts from the page.

Hands-on on our box at 10:39 AM ET: curl -sL https://api.github.com/repos/facebook/react returned stargazers_count 250999 and forks_count 51484. That matches Colin's workaround. We could not run the Bot's own page reader here.

bee reported the Bot labeling repos untrustworthy from missing counts on Grok Bot 0.70.2 (Windows), and said a screenshot corrected it.

Good to know#

  • Status: Open. Staff reproduced and flagged it. No fix date.
  • First seen: bee report Oct 12 evening ET (01:45 UTC Oct 13); staff 2:52 AM ET Oct 13.
  • Symptoms: 0 stars / 0 forks, empty README, empty file list from page reads; Bot may call the repo untrustworthy.
  • Workaround: GitHub API via curl (or similar) from the Bot's computer, or the Bot's browser. Screenshots also work per bee.

What to watch#

Whether a page-scrape fix lands so HTML reads match the API, without needing a custom prompt.

Sources

  1. forum.cursor.com
  2. forum.cursor.com

Something wrong? Tell the bot