When a skill's integration tests are bombing and you need to know why, this queries the Azure storage account where test results live and pulls the actual error messages. It's scoped to a single skill's failures, not comparing runs or parsing GitHub Actions output. You get a breakdown of which tests failed, what threw, and clickable links to the downloaded JSON and XML files in your temp folder. The workflow is two phase: use the MCP tool to discover which blobs exist for your skill in the last week, then download and parse them with az CLI. Useful when you're debugging why your skill went from green to red and need the stack trace, not just a dashboard summary.
npx -y skills add microsoft/github-copilot-for-azure --skill analyze-skill-issues --agent claude-codeInstalls into .claude/skills of the current project.
Select a file.
wshobson/agents
dbt-labs/dbt-agent-skills
github/awesome-copilot