/GRAFT Skill + Astra: THIS IS ABSOLUTELY CRAZY!
In this video, I’ll show you how Graft works with Verdent to help AI coding agents understand codebases more efficiently. We’ll build a structural code graph, investigate a permissions bug, trace function relationships, apply a fix, refresh the graph, and verify everything with tests. -- Key Takeaways: 🧠 Graft builds a queryable code graph containing functions, imports, and relationships across your project. 🔑 The structural graph works without an AI provider key or a separate database server. 🤖 Verdent can use the Graft CLI to investigate code, inspect files, make edits, and run tests. 🔍 Graft queries and skeleton views help agents find implementations, inspect APIs, and trace function callers. 🐛 In the demo, Graft helps identify a missing export permission check in a document service. 🔄 Graft can detect stale graph data and automatically refresh changed files after an edit. ✅ The final tests confirm that viewers can still read documents but cannot export them without permission. 📊 Graft also provides an interactive visualization for exploring dependencies across a codebase. ⚠️ Static analysis and token-saving estimates have limitations, so graph results should still be verified with source inspection and tests. 👍 Overall, Graft provides useful structural context that can make AI-assisted coding workflows more focused and efficient.
