KIT formally verifies ~1,000 LangGraph functions, surfaces state-loss and auth bugs
LingmingZhang · x · 2026-10-23
Grigore Rosu's team used KIT, a formal-verification tool, to verify 1,000 functions in LangGraph and reported the bug classes found to maintainers (one fix already landed).
The proofs surfaced several bug classes:
- state silently lost after one failed save
- a rejected action still going through
- user data stored unencrypted
- an auth check that could be skipped
KIT's pitch is "prove your code matches your intent": it extracts a formal spec from code (built on the K framework) and delivers math-backed verdicts via a Prover, rather than fuzzy LLM judgment. It currently supports Python and Solidity/EVM, with Rust, C, and JavaScript on the way. It ships as a Claude Code plugin and Codex integration—letting AI coding agents generate and verify specs automatically, an interesting new "verification layer" for agent workflows.
More from coding & agent
- LLMs make code so cheap that recursive rewrites could make verified software tractable — jd_pressman · 2026-10-23
- Hugging Face hands OpenEnv RL environments project to Linux Foundation for neutral governance — lmoroney · 2026-10-23
- Theo writes the definitive explainer on what an AI coding "harness" actually is — pvncher · 2026-10-23
- Conga launches MCP Server letting Claude and Copilot query live contract data — shashib · 2026-10-23
- LLMs make code cheap enough that recursive rewrites become tractable, argues researcher — jd_pressman · 2026-10-23
- Skip the rules file: point your coding agent at Clean Code instead — Al_Grigor · 2026-10-23