← back to all work
Case 01 · 2026

BugAgent

BugAgent preview

An autonomous LLM system on LangGraph that orchestrates four specialized agents, Scout (intel ingestion), Brain (rules, PRD, and hypotheses), Hacker (browser, Nuclei, and ZAP probes), and Diplomat (reports and disclosure email), from scope ingestion to submission-ready vulnerability reports backed by real proof-of-concept evidence. A ChromaDB-backed RAG knowledge base feeds the pipeline, while hardcoded scope, method, and rate guardrails plus mandatory human approval keep it strictly ethical.

Multi-agentLangGraphRAG
Next caseFuelNetEnv