Graph Traversal Agent for Code Debugging
summary
An agent uses Neo4j Graph DB and a BL graph Python package to map repositories into a code graph, then traverses it with simple tools (node-by-id and keyword search) to debug, optimize, and find vulnerabilities. Demo builds a graph of the repo with 210 nodes and 369 edges, the agent leaves breadcrumbs (short node summaries and embeddings) and the LM+graph traversal successfully answers questions and pinpoints functions, Sentry errors, and PR issues.
AI-written summary · about
chapters (8)
- 0:00Introduction and agenda
- 1:15Speaker backgrounds
- 2:40Overview of the Blar platform
- 4:26Structuring code as a graph
- 5:30Contextualizing code with data
- 6:55The importance of context
- 8:34Platform and technical demo
- 14:21Q&A and closing remarks
related talks
30mTracing the LLM Behaviour: Debugging and Evolving Agent Workflows with Neo4j
24mGDS Agent for Graph Algorithmic Reasoning
29mTransparent Multi-Agent AI
28mBeyond Vibe Coding: AI Agents for Technical Debt Detection
47mExploring a complex codebase with AI
01h 17mGraph Database and Retrieval Augmented Generation
30mLive-Coding Graph-Native Agents: Neo4j AuraDB + Microsoft Agent Framework in Action
01h 00mGraph Mastery to AI