AI ready repo

[DonJayamanne]

• Merge plan.bug.prompt.md and plan.feature.prompt.md into one
• Create triage.prompt.md to help traige issues
• Merge/remove/update analyze.bug.prompt.md, should this just be research.prompt.md (i.e. just a subset of what we already do in plan.prompt.md
• Testing of new unit tests don't work. When model creates a new unit test it doesn't run, as its not yet discovered. Need to open test explorer and refresh list of tests, then works.
  • Repro steps
  • Create a new test file
  • Open test file
  • Select command Run Tests in current file & you get a message that there are no tests
• Do we need a triage prompt, or should that be a whole separate process or notebook to perform the analysis. Like what Michael has done. The goal here to come up with the labels bug, feature request, not planned, as designed, etc. I.e. we need the model to analyse the issue and dig into the code and figure out the details.
• update.instructions.prompt.md At the end of the above process, should we ask model to add/update instructions. Or should the model capture all findings incrementally through the /plan & /execute phases and finally collate that information and add/update the instructions? Latter seems better as more information/context will be preserved & its just a matter of summarizing that at the end.
• Model looks to run tests using npm test:* (sometimes we can see copilot looking for tests)
• Copilot needs the ability to run integration tests, not just unit test (its too time consuming and slow to go back to copilot telling it that the tests have failed and provide the details of the failure)
• Copilot Coding Agent and Models try running specific tests with the grep argument
• New Integration Tests don't show up in test runner hence unable to run them in Agent Mode Errors in discovering vscode tests #16916