Use generative AI to help fill in request data (and other fixes)

[brettimus]

PR that makes the genAi features public, and fixes a few smaller bugs:

• FP-3842
• FP-3867
• FP-3861
• FP-3847
• Sometimes log entries said console.lognull
• Render 20 items per page in requests table by default
• Don't use relative timestamps in the requests table

---

For now, will require inputting your own OpenAI api key.

(See: "Coming Later" section for plans to support other models, as well as local models.)

Testing Plan

This branch is published as npx @fiberplane/studio@beta if you want to test end-to-end. Otherwise, feel free to test locally

• Go to Settings
  • Enable AI features in Settings
  • Use a company API key for openai (can also use a personal one, it will only use pennies)
  • Test with gpt-4o or gpt-4-turbo (gpt-3.5-turbo does not work with our prompt for some reason)
  • Click “Save”
• Go to Routes
  • Generate query parameters
  • Generate headers
  • Generate body
• Select Hostile testing persona
  • Generate query parameters
  • Generate headers
  • Generate body
• For one of the hostile scenarios, select “Create Test” in the bottom of the response panel
  • Copy-paste into Copilot or Cursor or ChatGPT or Claude

Coming Later

• Anthropic / Claude support
• Custom testing personalities
• Optional AI Summaries of errors
• Vary responses more based on history of generated requests

Coming Even Later

• Local models for query parameter generation