Sandevh/stu 307 generate tests with an llm

[izi-on]

• cleaned up code +
  GENERATOR PANEL:
• generate tests with a given prompt

[linear]

STU-307 Generate tests with an LLM

[vercel]

The latest updates on your projects. Learn more about Vercel for Git ↗︎

Name	Status	Preview	Comments	Updated (UTC)
docs	✅ Ready (Inspect)	Visit Preview	💬 Add feedback	Mar 5, 2024 10:11pm

[LatentDream]

Backend

How do we handle inference? Do we pay for it? Not sure if we can package our key with the desktop app due to security reasons. We won't be able to update the key if something happens.

Or should we let the user add their own OpenAI key? Set this key the same way as the cloud one?

Any other solutions?

---

Frontend:

• The Run Test Sequence button should always be visible, in my opinion. We could move it to where the Play button is on the flowchart.

• A toast.promise would be great so the user knows the request is being processed in the background.

• Now that the import and all are not always present, it's easy to confuse the following:
  

• For hardware, it completely hallucinate function (with tm_devices as example). We need to do some tests on how we can use other models, give it more context, fine-tune, etc. :)

  • I have some idea that I want to test when the export feature is done for that :)

• Right-click to consult code in the generated panel doesn't work.

  • shoud we do a mouse over to check the code directly ?
    

• I haven't went through all the frontend code wet :) But the UI is clean !!

[LatentDream]

key should be as an ENV variable, never hard coded since it can easily be forgotten