July 28, 2026
Proofs, panic, and parser gossip
Show HN: Lean4 Datalog DSL Based on Google Zanzibar for AI Projects
Proof nerds made a project tracker and the comments instantly turned into Lean 101 chaos
TLDR: ZIL Lean is a new Lean 4 tool that maps how code, requirements, tests, and documents connect so teams and AI can query the whole project more easily. But the comment spotlight went to confused readers asking the big basic question: is Lean a proof tool, a programming language, or both?
A Show HN post about ZIL Lean should have been a neat little launch story: someone built a tiny language inside Lean 4 that tracks how parts of a software project connect, like which code fulfills a requirement, which test backs it up, and what breaks if you change something. The pitch is that humans, tools, and even AI helpers could all look at the same relationship map and ask useful questions. Very organized, very tidy, very "what if your codebase had a family tree."
But the real action was in the replies, where the vibe instantly shifted from "cool tool" to "wait... what even is Lean?" One commenter, sterlind, basically became the audience avatar and unleashed a glorious spiral of confusion: is Lean a proof tool, a programming language, both, neither, or some wizard school where "tactics" secretly count as functions? That uncertainty became the story. Instead of arguing over the product alone, the thread turned into a mini identity crisis about Lean itself.
And honestly, that was the entertainment: not furious outrage, but deliciously nerdy bewilderment. The hottest take wasn't "this is bad" or "this changes everything"; it was the timeless comment-section classic of someone bravely asking the question everyone else was pretending to understand. In a sea of relation maps and Google-inspired access rules, the funniest subplot was simple: the community got distracted and turned the launch into an accidental Explain Like I'm Five for theorem-proving software.
Key Points
- •ZIL Lean is a Lean 4-based relational DSL for representing named objects, their relationships, and inference rules.
- •The system records connections among code declarations, requirements, documents, tests, tasks, dependencies, and proofs.
- •Its data model is influenced by Google Zanzibar’s tuple-oriented authorization system, including tuple references to other relations through usersets.
- •ZIL uses Datalog-style Horn rules to derive new relations, such as propagating document viewer access from group membership.
- •The article provides a software project example linking a parser, a normalization pass, and a theorem through `implements`, `dependsOn`, and `validates` relations.