Replace jest integration tests with vally eval suites

[JasonYeMSFT]

Description

The existing jest integration tests are written to manually call into the agent runner and assert the results. We have decided to migrate them to use vally to take advantage of its parallelization and rich set of graders. Unfortunately, vally doesn't support several critical features that our tests use, such as

• early termination
• follow up
• system prompt modification
• screenshot

In addition, it doesn't publish the necessary data that our integration test dashboard consumes for viewing the test results.

This PR creates a vally custom executor to support the missing features and publish all necessary data that the integration test dashboard depends on. The underlying agent runner used by both the existing integration tests and the vally custom executor is patched to achieve temporary backward compatibility.

Together with the infrastructure change to enable migration, this PR migrated existing vally eval suites to use the custom executor and tested on them to make sure the extended features and result data publishing work as expected. Some patches were made in the agent runner to achieve backward combability, supporting both Jest integration tests and vally eval suites with shared agent runner.

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

#1821
https://github.com/microsoft/evaluate/issues/276

[Copilot]

Pull request overview

This PR migrates Copilot SDK-based integration testing from bespoke Jest integration tests toward Vally eval suites by introducing a custom Vally executor, adding stimulus validation tooling, and patching the shared agent runner to support both legacy Jest tests and the new Vally path.

Changes:

• Added a custom Vally executor (integration-test-agent-runner) plus tag helpers to emulate missing Vally features (early termination, follow-ups, system prompt changes, screenshots) and to generate dashboard-compatible artifacts.
• Updated the shared agent runner APIs and updated Jest integration tests/scripts to use the new useAgentRunner({ isTest, useJest, ... }) config shape.
• Added a scripts CLI for validating eval stimuli metadata and updated CI workflow + eval specs to use the new executor.

Show a summary per file

File	Description
tests/vally/vally-executor.ts	New custom Vally executor that runs the Copilot SDK agent runner and attempts to post-process Vally outputs for the dashboard.
tests/vally/tag-helpers.ts	New helpers to parse stimulus tags into early-termination/follow-up/system-prompt/screenshot behaviors.
tests/utils/evaluate.ts	Updates import to include .ts extension for ESM/TS extension compatibility.
tests/utils/agent-runner.ts	Extends agent runner config (model/systemPrompt/workspace) and adds a new useAgentRunner configuration shape for Jest vs Vally usage.
tests/tsconfig.json	Enables TS extension imports and configures TS for typecheck-only (noEmit).
tests/scripts/generate-test-reports.ts	Updates to new useAgentRunner({ ... }) signature; forces explicit process exit on success/failure.
tests/README.md	Documents ongoing migration to Vally and points to executor and eval docs.
tests/package-lock.json	Bumps nested @github/copilot dependency under @microsoft/vally from 1.0.44 → 1.0.45.
tests/microsoft-foundry/quota/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/models/deploy/deploy-model/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/models/deploy/deploy-model-optimal-region/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/models/deploy/customize-deployment/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/models/deploy/capacity/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/troubleshoot/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/trace/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/observe/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/invoke/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/deploy/integration.test.ts	Updates useAgentRunner call to new config object.
tests/microsoft-foundry/foundry-agent/create/integration.test.ts	Updates useAgentRunner call to new config object.
tests/entra-app-registration/integration.test.ts	Updates useAgentRunner call to new config object.
tests/entra-agent-id/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-validate/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-upgrade/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-storage/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-resource-visualizer/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-resource-lookup/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-reliability/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-rbac/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-quotas/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-prepare/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-messaging/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-kusto/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-kubernetes/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-kubernetes/azure-kubernetes-automatic-readiness/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-hosted-copilot-sdk/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-enterprise-infra-planner/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-diagnostics/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-deploy/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-deploy/avm/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-cost/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-compute/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-compliance/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-cloud-migrate/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-aigateway/integration.test.ts	Updates useAgentRunner call to new config object.
tests/azure-ai/integration.test.ts	Updates useAgentRunner call to new config object.
tests/appinsights-instrumentation/integration.test.ts	Updates useAgentRunner call to new config object.
tests/airunway-aks-setup/integration.test.ts	Updates useAgentRunner call to new config object.
tests/_template/integration.test.ts	Updates template to use the new useAgentRunner config object.
scripts/src/vally/validate-stimulus.ts	New validator that enforces required tags/metadata conventions across evals/**/eval.yaml.
scripts/src/vally/cli.ts	New npm run vally ... CLI entrypoint for Vally-related repo tooling.
scripts/package.json	Adds vally script to run the new CLI via tsx.
evals/microsoft-foundry/eval.yaml	Switches executor to integration-test-agent-runner.
evals/entra-app-registration/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-upgrade/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-storage/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-prepare/eval.yaml	Switches executor and adds missing per-stimulus cost: llm tags.
evals/azure-kusto/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-hosted-copilot-sdk/eval.yaml	Switches executor and adds missing per-stimulus cost: llm tags.
evals/azure-enterprise-infra-planner/eval.yaml	Switches executor and adds missing per-stimulus cost: llm tags.
evals/azure-diagnostics/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-deploy/eval.yaml	Switches executor, adds missing cost: llm tags, and adds a new stimulus migrated from Jest.
evals/azure-compliance/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-aigateway/eval.yaml	Switches executor to integration-test-agent-runner.
evals/azure-ai/eval.yaml	Switches executor and adds early-terminate tags; removes completed graders from the shown stimuli.
.github/workflows/eval.yml	Adds stimulus validation step; adjusts fork gating to step-level for eval execution; formatting updates.
.github/skills/vally-eval/SKILL.md	Adds a repo-local skill describing how to author/validate/run Vally eval suites and use the custom executor.

Copilot's findings

Files not reviewed (1)

• tests/package-lock.json: Language not supported

• Files reviewed: 67/68 changed files
• Comments generated: 9