Skyramp

We're making enterprise-grade, deterministic testing accessible via AI coding assistants and agentic IDEs. Get beta access in @cursor_ai or @code today!

read the blog post skyramp.dev/blog/skyramp-a…

We're making enterprise-grade, deterministic testing accessible via AI coding assistants and agentic IDEs. Get beta access in @cursor_ai or @code today!

Deterministic testing in Cursor. The @skyramp_dev Agentic Experience is now available in @cursor_ai. Get beta access by installing the Skyramp MCP Server in Cursor.

Deterministic Testing in Visual Studio Code. @skyramp_dev Agentic Experience is now available in VS Code @code Get beta access by installing from the Extensions Marketplace.

@aidenybai Trust your coding agent? Sure, just don’t trust it with testing yet. That’s one job we’ve learned is too critical to outsource blindly.

do you trust your coding agent? (e.g. Cursor) i've seen people: - instantly accept without reviewing - paste API keys, credit card info, PII - allow arbitrary SQL queries - ship without testing

@therealdanvega Postman is solid. Lately, we’ve been leaning on a local-first generator that spits out ready-to-run API tests from specs/flows in seconds. Hard to go back.

What API Testing tools are you using today? Postman/Insomnia/Something else? I know there are some great tools built into IDE's but I'm just curious what tools everyone is using these days?

@kihaki @JorgeCastilloPr Testing in prod can work. But skipping tests entirely means you're outsourcing confidence to your error logs. Move fast. But have a net. Preferably one you don't have to rewrite every sprint.

@JorgeCastilloPr Testing in Production is completely feasible and the best way to iterate quickly, writing unit tests is a waste of time in 95% of cases

What's a opinion about Android development that would have people doing this?

@KevinNaughtonJr Finally, a clean 3000-line function and one glorious test case to rule them all 😌 Jokes aside, maybe the real fix isn't fewer functions - it's better ways to generate tests?

your entire application should live in one function so that you only need to write unit tests for a single method

@growing_daniel @tekbog Ah yes, the classic “hope your memory is better than your CI” testing strategy. Feels like this could be… automated?

@tekbog Unit tests are trash but you should keep a mental map of everything your e2e tests touch and make sure you have full coverage by proxy

after years u figure out that clicking around is much better than writing unit tests btw

@jerkeyray Most devs skip testing for personal projects. Not because it's unimportant. Because it's annoying. What if generating tests actually felt as smooth as building?

is there even a point to be writing unit tests for personal projects? like can't you just see if the thing is working or not?

@n0w00j The real high is not just getting tests to pass. It's knowing they'll stay passing when your code changes.

alcohol is cool, but have you ever had your unit tests pass on the first try? me neither. pass the bottle.

@CodeFather10 You code fast. You ship fast. But tests? They always slow you down. You patch one thing, and now three test files are on fire. What if writing and maintaining them wasn’t so painful in the first place?

Writing unit tests feels like paying taxes. You don’t want to, but you know it’ll hurt more later if you don’t.

@slimjimmy 31. If the unit tests pass, the app works. Passing tests don’t mean working software - they just mean the tests passed. You can have a green checkmark and a broken user flow.

falsehoods programmers believe about unit tests: 1. unit tests are all you need 2. unit tests mean you don't need QAs 3. unit tests mean you don't need to manually test 4. unit tests are the most important test 5. unit tests are the 2nd most important test 6. unit tests find defects 7. you should never delete unit tests 8. you should never change unit tests 9. integration tests = unit tests 10. e2e tests = unit tests 11. coverage is a useful metric 12. 100% coverage is a target 13. 80% coverage is "good" 14. unit tests lead to good design 15. unit tests should have one assertion 16. unit tests require mocking frameworks 17. mocking frameworks are useful 18. unit tests that only test the interface don't fix implementation in place 19. unit tests mean you don't need types 20. types mean you don't need unit tests 21. unit tests = TDD 22. unit tests are crucial for refactoring 23. unit tests are crucial for CI/CD 24. unit tests only test "units" 25. unit tests are good documentation 26. bad unit tests are better than no unit tests 27. no unit tests are better than good unit tests 28. unit tests enable code reuse 29. unit tests are useful in prototyping 30. TDD by Example is a book on testing

@JonTuckerUSA This is what the AI era looks like: You bring the vision. Your dev partner brings the speed. But if you’re not thinking about how to test that velocity, you’re just shipping bugs faster.

I can’t code but from scratch - but I’ve gone DEEP into learning software development and AI coding. Pure vibe coding an app doesn’t work. But collaborating with a solid dev that is as all in on AI coding as I am is mind blowingly fast. This is the way 🔥🚀

@unclebobmartin @clar1k This nails it. Integration tests tell you if the system does the right thing. Unit tests tell you why it does the right thing. You need both to build systems that last and don’t break quietly.

@clar1k Good. Integration and acceptance tests are good for demonstrating that the system conforms to the business requirements. Unit tests demonstrate that the system conforms to all the engineering requirements that the business does not understand.

we do code for business not just to see 100% on test coverage

@samim The problem isn’t just writing tests. It’s keeping them trustworthy as your system evolves. LLMs help you move fast. Your tests should too. Self-maintaining test suites aren’t a nice-to-have anymore.

Pro Tip: Use your LLM not just to write code, but to stress-test it. Auto-generate exhaustive tests—unit, integration, edge—and rerun them frequently. Trust comes from verification.

@penberg Reproducing flaky bugs manually is brutal. What if tests weren’t just snapshots of known states, but self-maintaining blueprints of expected behavior? That’s where testing is headed.

A big problem with debugging bugs is that the more interesting ones are hard to reproduce. That's why people build deterministic simulators to test their software or use tools like Antithesis. But why are interesting bugs so hard to trigger? 1/

@tom_doerr Rerolling LLM code works great - until the test suite starts failing on old assumptions. Code generation is fast now. Your tests should be just as adaptable.

Thinking more and more that you shouldn't debug LLM code. Generate edits, and if tests fail, undo and regenerate. Optionally update your prompt.

@Kiptanui_boazo Every frontend dev has war stories from untested APIs. One minute you're building UI, the next you're reverse-engineering a backend you didn’t write. A little API testing goes a long way.

I believe I speak for all frontend devs when I say we've all at some point in the hands of backed devs who never test their APIs A lot of valuable time is lost debugging

@kettanaito Love this. Debugging is part of the test lifecycle. The real skill isn’t just writing tests, it’s knowing how to learn from the ones that break.

I'm going to have another testing workshop that has an entire section dedicated to Debugging. Such an overlooked topic. What should you do when test fail? Because that's not an "if", that's a "when". You write tests for them to fail. How to approach it when they do?