How do you _guarantee_ your agent's code is correct? You prove it. Introducing Proof Driven Development Kit (PDD Kit) from Microsoft Research. PDD Kit helps turn your plan into a formal spec, then proves the agent's code meets that spec for every input. No vibe verification. PDD Kit is... integrated with GitHub Next's Chopin, so you can plan as a team. PDD Kit takes an app's login flow from intent to a verified pull request: It discusses your code, infers your intent as it goes, and proposes a formal spec in Verus, a proof language for Rust. Once you approve the spec, the Verus verifier automatically proves the code meets it. Where proof isn't possible, it falls back to empirical verification with tests and computer-use agents (CUAs). It works on Rust today. No new language, no change in how you produce the underlying code, making formal methods more accessible. The goal is a verified software factory, where engineers review specifications instead of code. Come see it in action at GitHub Universe, Oct 28–29. Swing by the Speakeasy and ask for GitHub Next!show more

@MSFTResearch sounds like a great idea, good luck!

@MSFTResearch proving against a formal spec still leaves the spec to review, but that is a lot less to read than the code

@MSFTResearch Does it Open-source?

@MSFTResearch A verifier can prove code meets its formal properties, but it cannot prove those properties fully capture what the user actually intended

@MSFTResearch @VictorTaelin

@MSFTResearch @VictorTaelin

@MSFTResearch Your honor, this computer use agent clearly exonerates my client

@MSFTResearch A coding agent saying “trust me” is a tiny Kremlin spokesman. PDD Kit turns intent into a formal spec, then proves Rust code meets it for every input. Less vibes, more receipts.

@MSFTResearch haven't read but yes, this is what I want

@MSFTResearch no vibe verification is going on a shirt

@MSFTResearch So just test driven development ?

@MSFTResearch @VictorTaelin they're stealing your work

@MSFTResearch RIP to the other PDD Kit (Product-Driven Development)

@MSFTResearch @grok can I used this in gitlab too?

@MSFTResearch @bot send me the summary and how it is related with my mental modal

@MSFTResearch make the proof-to-test fallback visible in the pull request. otherwise reviewers can mistake a passing test suite for proof.

@MSFTResearch So basically a really deep integration with a test library in github?

@MSFTResearch A cheap step before full proofs: property-based tests on the few functions where a bug costs real money, like pricing, permissions, and refunds. Hypothesis or fast-check runs thousands of generated inputs against the invariant, and the shrunk failing case is the bug report.

@MSFTResearch i tried to prove my code was correct, but my agent said 'if it works, it's good enough'

@MSFTResearch Proving the code matches a spec still leaves the tougher question: does the spec actually cover the real-world failures and constraints the system will hit?

@MSFTResearch @bot can you remind me of this?

@MSFTResearch @grok could this PDD be a simple pr review skill ?

@MSFTResearch how long does Verus take to prove that login flow, seconds or minutes?

@MSFTResearch @G_Programming

@MSFTResearch Reviewing the spec instead of the diff is a big shift, and falling back to tests where proof isn't possible feels honest. Any plans beyond Rust, or is Verus the main reason it starts there?

@MSFTResearch Thoughts on this @VictorTaelin ?

@MSFTResearch Is this going to be a programming language agnostic specification?

@MSFTResearch Is there a repo?

@MSFTResearch @grok I didn't read any of this. Does it use an agent or LLM to verify or is it deterministic?

@MSFTResearch @bot verifica se isso aqui é útil, por favor