A collection of property-based tests for open-source projects, in every language Hegel supports. We aim for these to be high quality tests, but this is definitely not currently fully achieved.
Important
We have tried to be careful to not include any projects that have said that AI contributions are unwelcome, but if we've missed something or you would like us to remove your project from here for this or any other reason, please also let us know and we would be happy to do so.
These tests are fully LLM generated, and are primarily for our own evaluation of Hegel. Where possible, we report bugs upstream, but in order to be respectful of maintainer time we don't do that without a human review step. As a result, there are a large number of bugs here that have not yet been reported.
If your project is in the zoo and you would like our help triaging any bugs we found in it, or in integrating Hegel into your testing, please get in touch and we'd be very happy to help you with any of this.
Each target — one upstream project in one language — lives at targets/<lang>/<name>/:
target.toml: upstream URL, pinned base commit, Hegel version, how to run, and the tests that are expected to fail because of a known, still-open bug;hegel.patch: the tests, as agit apply-able diff against the base commit, plus the one-line dev-dependency on Hegel;README.md: what is tested, with which oracles, and what deliberately is not;bugs.toml: every bug the tests found, with its status and a per-version history. Note that these are not necessarily validated by a human yet, so we don't guarantee that any "bugs" listed there are genuinely bugs. However we've generally found the reliability of the agent reports pretty good so most of them probably are.
zoo.toml pins the Hegel library version per language. TROPHIES.md is generated from all
the bugs.toml files. DESIGN.md explains the choices.
HACKING.md says how to run the tests and how a target for each language is put together.
STYLE.md says how the tests should be written (idiomatic Hegel: generators as values,
combinators over hand-rolled choice), what a survey of the existing tests found wrong, and
what Hegel itself could add to make good tests easier to write.