Run comparison integration tests without skills

[JasonYeMSFT]

Description

Extend the integration test workflow to run comparison tests on Sunday. Unlike regular integration tests, these comparison tests don't load any skills. They are intended for comparing the behavior of the agent with/without our skills to help us guage if our skills are helpful and if they may cause the agent to use significantly more tokens.

Currently, the cron job config only runs azure-deploy tests for the comparison tests.

Checklist

• Tests pass locally (cd tests && npm test)
• If modifying skill descriptions: verified routing correctness with integration tests (npm run test:skills:integration -- <skill>)
• If modifying skill USE FOR / DO NOT USE FOR / PREFER OVER clauses: confirmed no routing regressions for competing skills

Related Issues

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

Extends the integration test workflows to support “comparison” runs that execute without loading skills (via NO_SKILLS), scheduled for Sundays.

Changes:

• Add NO_SKILLS gating to skip skill assertions and skill-invocation test suites during comparison runs.
• Add a Sunday schedule entry and propagate a no-skills input/env var through GitHub Actions workflows.
• Update PR validation logic to exclude the Sunday comparison schedule from “all skills covered” checks.

Show a summary per file

File	Description
tests/utils/evaluate.ts	Skips softCheckSkill when NO_SKILLS=true.
tests/skills.json	Adds a Sunday schedule entry intended for comparison runs.
tests/azure-deploy/integration.test.ts	Skips skill-invocation describe block when NO_SKILLS=true.
.github/workflows/test-azure-deploy.yml	Adds no-skills input and sets NO_SKILLS env for Jest runs.
.github/workflows/test-all-integration.yml	Adds Sunday cron and wires through no-skills resolution and env.
.github/workflows/pr.yml	Limits schedule coverage validation to weekday cron entries only.

Copilot's findings

• Files reviewed: 6/6 changed files
• Comments generated: 4

[jongio]

Clean separation - dedicated workflow keeps comparison runs from triggering weekday automation. The NO_SKILLS gating is straightforward and doesn't affect normal test runs. LGTM.