Google Reveals AVDH: Agentic Framework for Automated Vulnerability Discovery
rseroter · x · 2026-08-19
Google Cloud and Mandiant published a detailed blog post introducing the Agentic Vulnerability Discovery Harness (AVDH), an internal framework designed to automate the discovery of security vulnerabilities in large codebases.
Core Mechanism:
- Structures the analysis process and enforces skeptical validation steps.
- Injects domain-specific human expertise directly into the AI pipeline.
- Uses multi-agent orchestration to assist experts during proactive reviews, penetration tests, and incident response.
Real-World Results:
- AVDH has demonstrated a significant leap in efficacy over the past 10 months.
- In a recent incident response involving stolen corporate repositories, the harness identified over 100 true-positive critical vulnerabilities in just two days.
More from coding & agent
- Sentry CEO's screenshot: one Claude Code research run cost $101.90 and 243M tokens — zeeg · 2026-10-16
- One year of Tinker: Thinking Machines' API for owning your fine-tuned models — thinkymachines · 2026-10-16
- Architects vs gardeners: developer argues AI's plan-maxxing obsession hurts good work — nptacek · 2026-10-16
- OpenRouter launches Armor beta to block out-of-policy agent tool calls before execution — dfinke · 2026-10-16
- Meta's coding-optimized open-weight Muse Spark Open lands on Modal with 1M context — AAAzzam · 2026-10-16
- Google rewrites its SRE book for the 10th anniversary, adding agentic-era guidance — RealGeneKim · 2026-10-16