Steve Yegge: ~25% of adversarial agent reviews hold up, and he's keeping them
Steve_Yegge · x · 2026-11-04
Steve Yegge ran his agents over thousands of recent adversarial review records: roughly a quarter were blocked by the reviewer, and a meta-review found the issues raised by agentic reviewers were almost all legitimate.
Key details:
- He bans agents from filing P3 (low-priority) issues, which cuts noise significantly
- The quoted discussion warns that multi-round adversarial review tends to push agents off a cliff into over-engineering, with a trajectory toward formal verification → deterministic simulation → agents using the product like real users
- Yegge's verdict: it works for him and he's keeping the setup
More from coding & agent
- Oracle Launches Graph-Based Agent Memory Where New Facts Supersede Old Ones Automatically — CodeByPoonam · 2026-11-04
- Mem0 Article: Telling Your Agent What to Store — and What Not To — blaizedsouza · 2026-11-04
- Matt Pocock: In the Agent Era, Code 'Readability' Is a Nicety — Optimize for Verifiability Instead — mattpocockuk · 2026-11-04
- An Overview of 60 AI Code Reviewers — None Can Fully Replace Human Review — blaizedsouza · 2026-11-04
- How many MCP tools is too many? Rewriting descriptions as 'use this when' helps — AryanBulandi · 2026-11-04
- pr-agent, the original open-source AI PR reviewer, hits 14.5k stars and says it's not the Qodo free tier — The-PR-Agent · 2026-11-04