Zero-security dev's autonomous agent went from one-byte overflow to tcache poisoning in hours
Genaforvena · reddit · 2026-11-20
A developer with no exploit-development experience gave an autonomous agent (Mishe) newly disclosed vulnerabilities and no guidance. Within hours it reproduced a one-byte heap overflow in Ruby, built a working heap-overlap primitive, confirmed reachability via HTTP in Sinatra/Rails, and progressed to tcache poisoning (5/5 in-process successes, 3/3 clean controls). No full remote RCE yet. It ran on OpenCode Go fueled by €500 in credits; other agents are probing NGINX, Node.js and PHP. Everything is public under CC0, and the author asks exploit devs to reproduce it.
More from coding & agent
- Rhenium: a from-scratch Rust decompiler built for humans and AI agents — OwariDa · 2026-11-20
- RUC & Fudan release CoworkOS: agent runtime for long-horizon work with human edits, boosting benchmarks from 33.19 to 42.95 — jiqizhixin · 2026-11-20
- Starlight Arena offers $50k in prizes for talking AI agents into breaking their own rules — jim-jam-biscuit · 2026-11-20
- TrustedSec: AI agents make attacker attention cheap, rewriting offensive security economics — cyb3rops · 2026-11-20
- Apple job listing reveals an AI agent that browses, runs terminal commands and debugs code — mark_k · 2026-11-20
- Agno builds sandboxing for Claude Code/Codex agents with S3 checkpoint restore — pritisinghhhh · 2026-11-20