microsoft / microsoft/GitHub-Copilot-for-Azure
Add lightweight routing-only test to validate skill selection across models on every PR
- Dominant language
- Python
- Stars
- 250
- Forks
- 204
- Avg merge
- 1d 12h
- Merged PRs (30d)
- 67
Description
### Problem
Skill selection is only fully tested through slow, comprehensive test suites, limiting our ability to catch regressions early on every PR.
### Desired Solution
Add a light, fast test that validates routing-only/skill selection for all models, omitting full content quality checks. This can run on every PR without slowing down the workflow.
### Alternatives and Workarounds
Continue only using full test suites, or try to speed up those suites but risk missing skill selection bugs on PRs.
### Impact
Faster detection of routing/skill selection errors; prevent regressions from being merged due to lack of early feedback; allows full test suites to focus on deep correctness.
### Additional Context
Request originated in [Issue #1647](https://github.com/microsoft/GitHub-Copilot-for-Azure/issues/1647). Intended to complement, not replace, full suite coverage.
Contributor guide
Research direction
Locate the existing full skill-selection test suites and the pull-request workflow that runs them. Start by identifying how models are enumerated and how routing-only behavior can be isolated; done means a fast test covers all models on every PR without running full content-quality checks.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- testing
- Issue type
- Feature
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 55/100