Back to browse
GitHub Repository

Semantic diffs in plain English for Claude Code. See what changed in meaning, not just what lines moved. Detects breaking changes, affected dependents, test impact, and risk level. Zero LLM dependency.

2 starsTypeScript

VibeDiff – Blocks Claude Code from shipping breaking changes

by Boussettah·Mar 3, 2026·1 point·0 comments

AI Analysis

●●SolidSolve My ProblemShip It

Claude Code safety gate that blocks breaking changes; solves real foot-gun but limited to one AI tool.

Strengths
  • Semantic diff detection (removed exports, signature changes, affected dependents) beyond simple line-based diffs.
  • Three-hook architecture (PreToolUse, PostToolUse, Stop) is elegant — intercepts at right moments without requiring manual review.
  • Risk scoring (LOW/MEDIUM/HIGH/CRITICAL) with explicit reasoning makes Claude self-correct before shipping.
Weaknesses
  • Locked to Claude Code ecosystem — doesn't work with Cursor, Windsurf, or other IDE-based AI tools (major limitation).
  • Early-stage: zero stars, no evidence of real-world usage or whether the risk scoring actually prevents bugs in production.
Target Audience

Claude Code users, teams using AI-assisted development, developers prioritizing code safety.

Similar To

Conventional Commits hooks · Husky + semantic-release · ESLint pre-commit gates

Similar Projects

AI/ML●●●Banger

Intel Overdrive – Real-time tool intelligence for Claude Code

Your Claude agent finds MCP servers you didn't know existed before you build from scratch.

Solve My ProblemBig Brain
tijmendevries
202mo ago