OPEN TO A WORLD OF POSSIBILITIES
A home for
open harnesses.
Good things happen when we build together. We’re developing open-source harnesses, shared benchmarks, and a place for the community to improve them together.
Open source. Shared infrastructure. Yours to build on.
A library, one
harness at a time.
Useful tools take care and cultivation.
We’re building a collection of open-source harnesses: the tools, checks, and context that help agents do useful work.
Starting with Vera. Developing the shared foundations as we go. Leaving room for what we build together.
Explore VeraIntentionally a work in progress.The library will grow through real use, careful iteration, and community contributions.
01VeraApp store checksON THE SHELF
Find issues before you submit. Vera brings app store checks into your development workflow.
Get to know Vera02Shared foundationsHow the library comes togetherBUILDING
We’re working toward a common way to connect harnesses to agents, with shared conventions for running checks and understanding results.
IN DEVELOPMENT / THE UNIFIED API03Room for what’s nextShaped by the people who use it
Have a recurring problem, a useful tool, or a harness you’re already building? Help us decide what belongs here next.
Build with usFIG. 01 / USEFUL WORK, SHARED
Start with something real.
A surprising result, a missed issue, an edge case. Turn it into a small example that others can reproduce.
Every case makes
the benchmark
better.
Open source. Crowdsourced. Always improving.
Good benchmarks grow out of real experience. We’re building a shared body of test cases that anyone can inspect, challenge, and help improve.
- Open to inspection
- The cases, expected results, and reasoning belong alongside the code.
- Shaped by contributions
- New cases add perspectives a single team would miss.
- Built to be rerun
- A shared reference for testing changes and catching regressions.
What you know
can help us all.
The workshop is open.
A case from your work. A fix from your experience. A tool you wish existed. This gets better when more people have a hand in building it.
01Bring a real-world case
Show us where an existing check misses something. A small, reproducible example helps turn your experience into a useful test for everyone.
Share a case02Make a check better
Question an assumption, clarify an expected result, or improve a test. Careful review matters as much as writing the first version.
Talk through an improvement03Build something with us
Bring a harness you’re working on or a problem worth solving. We’re interested in useful foundations that others can build on, too.
Tell us what you’re buildingYou don’t need a finished project.
A good observation is a good start.
Built in public. Improved together.
Start a conversation