Devotd Github
NewAI agents

monotykamary/gdp-wired — a tool for proving state-rule invariants

monotykamary/gdp-wired is a tool that proves invariants of state rules in TypeScript and finds counterexamples with a GPU.

Repositorymonotykamary/gdp-wired ↗

monotykamary/gdp-wired is a tool that proves invariants of state rules in TypeScript and finds counterexamples with a GPU.

GitHub description: Laws about stateful rules, proven in TypeScript: a small trusted kernel, an untrusted prover, a Lean twin, and GPU searches for counterexamples and invariants.

stars
★ 15
Forks
1
Language
TypeScript
License
MIT
Read original source ↗View on GitHub ↗

monotykamary/gdp-wired is a TypeScript tool that works with rules that must hold even when the state changes, such as budget caps or rate limits. According to the repository description, it decides whether a rule holds and uses GPU search to find counterexamples, presenting an execution path where the rule breaks. It was created on October 16, 2026 and is released under the MIT license. No latest release is registered.

The tool focuses on automatically proving invariants of state rules, that is, conditions that must keep holding across every possible state change. It treats not only budgets and rate limits but also state transitions as rules, and supports state diagrams and XState models. It can also extract rules from code, and the repository includes examples from several domains along with comparison documents. Proof results are re-checked with Lean. Lean is a theorem prover that mechanically verifies whether a proof is correct. The description also presents a small trusted kernel that makes the judgment, an untrusted prover, and a Lean twin.

However, the published materials do not present figures for how long it takes to prove a rule, nor performance comparisons with other tools. The repository has 15 stars, and no latest release is registered either. So the structure combining GPU search with Lean re-verification can be confirmed, but it is difficult to judge from the currently available materials alone what range of rules it can handle, and how efficiently, in practical work.