# Atnia

- **Event:** [Built with Opus 4.7: a Claude Code hackathon](https://cerebralvalley.ai/e/built-with-4-7-hackathon)
- **When:** Apr 21 at 12:00 PM – Apr 27 at 2:00 AM (EDT)
- **Where:** Online
- **Team:** [Aritra Sarkhel](https://cerebralvalley.ai/u/Aritra)
- **GitHub:** https://github.com/Aritra003/stinkit.git
- **Demo video:** https://drive.google.com/file/d/1wtgEF3H_-W7mUY2UQKYZx23XrYSQ8wep/view?usp=sharing
- **Gallery:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/44

StinKit is an open-source code intelligence platform that indexes your codebase into a structural dependency graph and lets you query it nine different ways — through a CLI, a web dashboard, or an AI agent via MCP.

WHAT IT DOES (the three things nobody else has):

1. Architecture Diagram Comparison (stinkit see) — Upload a whiteboard photo, Lucidchart export, or any diagram. Opus 4.7's vision reads it at 3.75MP, extracts every service and connection, then compares it against your actual code graph. "Your diagram is 58% accurate. 2 phantom services. 3 missing connections. Here's the corrected version." No other tool on earth does this.

2. Codebase Q&A (stinkit ask) — "What would break if I removed the cache layer?" StinKit queries the structural graph for context, then Opus 4.7 explains the architecture using real file names and dependency counts. Not keyword search. Graph-augmented reasoning.

3. AI Refactoring Planner (stinkit plan) — "Replace sessions with JWT." StinKit analyzes every affected file, determines the safe change order (leaves before roots), and generates a sequenced migration plan with PR boundaries, effort estimates, and rollback points.

PLUS: Blast radius check in <2 seconds (fully offline), real-time file monitoring (stinkit watch), professional audit reports (8-section, Deloitte-grade with health scores and remediation roadmaps), incident forensics (stinkit trace), and an MCP server with 9 tools so Claude Code and Cursor can query your code graph autonomously.

THE NUMBERS: 167 source files. 407 tests passing. 17 API routes. 10 dashboard pages. 9 CLI commands. 9 MCP tools. 11 languages supported. MIT licensed. $0 to run.

WHAT MAKES IT DIFFERENT: Every output shows what StinKit knows AND what it doesn't — "Local completeness: 75% · 956 ambiguous local calls · Known blind spots: event emitters, DI containers." Other tools claim 100% accuracy. We show you the real number. That radical honesty is the product.

---

Markdown version of https://cerebralvalley.ai/e/built-with-4-7-hackathon/hackathon/gallery/44. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt