Jev Based Code Review
Jev-code-reviewer is a local tool and Chrome extension that classifies massive AI-generated pull requests into priority levels to reduce human review fatigue.
summary is enough The tool itself is a niche utility for massive AI PRs, but the comment thread offers a sharp debate on whether generating 230-file agent PRs is an engineering failure.
summary
Jev-code-reviewer attempts to mitigate the cognitive load of large agent-generated pull requests by automatically prioritizing changes into P0, P1, and P2 buckets. Running locally via a Node.js CLI and a Chrome extension, it translates raw diffs into natural language summaries and hides lower-priority code by default. However, the Hacker News community heavily pushed back on the premise, arguing that generating 230-file PRs reflects a broken development process rather than something that requires more AI tooling to fix.
key points
- The tool classifies diff hunks into priorities like P0 and P1 using local execution and a Chrome extension without sending code elsewhere except for required provider calls.
- Commenters overwhelmingly criticized the practice of generating massive 230-file agent pull requests, calling it an engineering smell rather than a problem solved by adding more AI.
- Critics questioned the necessity of using Jev instead of traditional LLMs or custom classifiers for parsing and understanding code diffs.
who should read it
Developers experimenting with AI coding agents who struggle to review large, automated pull requests.
discussion hostile
consensus
Commenters broadly agreed that generating 230-file pull requests is a process failure and an anti-pattern. Most expressed fatigue with AI slop and argued that agents should be prompted to break changes into smaller, manageable chunks instead of relying on more AI to review the mess.
pushback
A minority perspective acknowledged that in heavily degraded or legacy codebases, human comprehension is already lost, making automated AI workflows the only viable path forward.
I think your engineering process is fundamentally broken if you are generating PRs with 230 files changed so regularly that you need to bolt on more AI. You’re solving the wrong problem.
Pinpoints the core pushback against the tool's premise by arguing that massive agent-generated PRs indicate a broken workflow rather than a need for more AI tooling.
Generated by an LLM from the article and thread. Check the sources before relying on it.