Show HN: Unravel-I built a deterministic debugging layer for AI coding agents

Category: devtools

Tags: ast-analysis, ai-debugging, mcp-server, deterministic, hallucination-prevention

Score: 6.8/10 (Innovation: 8, Technical: 7, Documentation: 6, Utility: 6)

Unravel is a deterministic debugging layer for AI coding agents that uses AST analysis to provide verified structural facts about code, preventing hallucinations in AI-generated debugging. It enforces a 'Sandwich Architecture' where deterministic evidence and verification surround AI reasoning, and it has demonstrated real-world utility by finding bugs in cal.com and tldraw. The project uniquely combines tree-sitter AST parsing, MCP protocol integration, and anti-sycophancy guardrails to force AI agents to debug based on code facts rather than pattern-matching symptoms.

Target audience: backend devs, ai-agent users, devops

Repository: https://github.com/EruditeCoder108/unravelai · JavaScript · NOASSERTION · 17 stars

View on Hacker News