Show HN: Agent-roast – send your coding agent back when it weakens a test

Category: devtools

Tags: ai-agents, test-quality, linter, ci-cd, git-hooks, static-analysis

Score: 7.3/10 (Innovation: 7, Technical: 7, Documentation: 9, Utility: 6)

agent-roast is a deterministic, offline linter that diffs test and CI changes to catch weakened checks — loosened assertions, skipped tests, lowered coverage thresholds, and CI steps allowed to fail. It integrates as agent hooks for Claude Code, Codex, and Cursor, as a Git hook, and as a GitHub Actions PR check, sending agents back once when they cheat. Its honest benchmark against 12,645 real commits and explicit acknowledgment of false-positive rates make it a credible, well-scoped tripwire for a real gap in AI-assisted development workflows.

Target audience: backend devs, devops, engineering managers, QA engineers

Repository: https://github.com/Nari-Ab/agent-roast · TypeScript · MIT · 9 stars

View on Hacker News