SIN-Code Proof of Correctness (POC)

Lightweight, practical formal verification for AI-generated code — property-based testing driven by function signatures and natural-language intent.

Part of the SIN-Code agent-engineering stack.

Why

Full theorem proving (TLA+, Lean) is too heavy for everyday agent loops. POC brings the spirit of formal methods to the practical layer: it infers properties (invariants) a function should satisfy, renders runnable Hypothesis tests, and verifies them at runtime — turning "looks right" into "checked against hundreds of inputs".

Features

• Property generator — infers candidate invariants (idempotence, monotonicity, length-preservation, sorted output, reversibility, purity, …) from a function's signature and name.
• Runtime verifier — executes properties with Hypothesis and reports PASS / FAIL / SKIPPED per property (no unsafe eval).
• Spec compiler — turns declarative pre/post-conditions, or a natural-language intent, into a property test.
• CLI (poc) for suggesting, verifying, and compiling specs.

Quickstart

pip install -e .
poc suggest mymodule.py my_function       # print suggested Hypothesis tests
poc verify mymodule.py my_function        # run runtime verification
poc from-intent "returns a non-negative, sorted list" my_function

Documentation

• INSTALL.md
• docs/USAGE.md
• docs/CONFIGURATION.md
• CONTRIBUTING.md
• CHANGELOG.md

License

MIT — see LICENSE.