Awesome Reviewers

Verification · Code review · Agentic exploration

Deep exploration

Focused sub-agents that investigate one concern each.

How to build it

  1. Start one sub-agent per concern from the survey.
  2. Give each sub-agent read, search and run tools for its own concern only.
  3. Merge the sub-agent findings into one list with the evidence for each.

Signals in a repository

  • A graph or workflow definition that fans out one branch per concern.
  • Review comments that cite evidence from several files for one finding.
  • Stored traces of each sub-agent run kept next to its findings.

In practice

No scanned repository runs a tool for this check. The steps above are the only guidance.