Agents Lookup

Use cases / Decision models (Jev)

Screen new GitHub repositories for ones worth a look

Crawl recently created repositories in a topic and have Jev decide from each README whether it deserves a person's time.

DifficultyBeginner
Setup time2-3 hours
CategoryDecision models (Jev)

Author's figure: Author's figure: 2,000 repositories read with 99.3% thrown out (Paone). Reported by the builder, not measured by us.

The use case

Developer relations staff, investors and engineers tracking a fast-moving area face thousands of new repositories a month, most of them empty or copies. A crawler pulls each new repository's README and metadata, and Jev gives a probability that it deserves a closer look, given what we care about. Nearly all get discarded, and the survivors go into a short weekly list. Someone still opens each listed repository, reads the code and checks the licence before mentioning, adopting or funding anything.

Example workflow

  1. Query the GitHub search API for repositories created in the last 30 days on a topic
  2. Fetch each README and basic metadata
  3. Ask Jev a Noul question on whether it is worth reviewing
  4. Write survivors to a Notion or Airtable list
  5. A person reviews code and licence before acting

Tools

Risks and guardrails

Sources

Written by Agents Lookup. Credits and official pages for the tools this use case names: