Upgrade eval workflow: pin latest action SHAs, global tool installs, COPILOT_GITHUB_TOKEN

[Copilot]

Modernizes .github/workflows/eval.yml to use the latest GitHub Actions, install vally and the Copilot CLI as global tools, and switch to COPILOT_GITHUB_TOKEN for evals.

Changes

• Upgraded actions to latest versions, SHA-pinned per security policy:
  • actions/checkout → v7.0.0
  • actions/setup-node → v6.4.0
  • actions/upload-artifact → v7.0.1
• Added "Install tools" step — installs @microsoft/vally-cli@0.6.0 and @github/copilot@1.0.65 globally so downstream steps invoke them directly
• Replaced npx -y @microsoft/vally-cli@0.6.0 ... with vally ... in Lint and Run evals steps
• Switched env var in Run evals from GITHUB_TOKEN → COPILOT_GITHUB_TOKEN: ${{ secrets.COPILOT_GITHUB_TOKEN }}

[Copilot]

Pull request overview

Adds Vally-based evaluation specs and fixtures for selected plugin skills and wires them into CI via a dedicated pull-request workflow, so skill behavior can be continuously validated as part of the plugin marketplace repo.

Changes:

• Added .vally.yaml defining a pr suite that runs linting + security eval specs.
• Added Vally eval specs + fixtures for check-spelling, lint-markdown, and pin-github-actions.
• Added a dedicated pull_request GitHub Actions workflow to run Vally lint/evals and upload vally-results.

Show a summary per file

File	Description
evals/security/fixtures/pin-github-actions/workflow.yml	Adds a workflow fixture used to validate action pinning behavior.
evals/security/eval.yaml	Defines the security plugin eval stimulus and graders for pin-github-actions.
evals/linting/fixtures/lint-markdown/input.md	Adds a markdown fixture with intentionally inconsistent bullet styles.
evals/linting/fixtures/check-spelling/input.md	Adds a spelling fixture containing a known typo and a project-specific word.
evals/linting/eval.yaml	Defines linting plugin eval stimuli and graders for check-spelling and lint-markdown.
.vally.yaml	Introduces the Vally project config and pr suite wiring to the eval specs.
.github/workflows/vally-pr-evals.yml	Adds a PR workflow that runs Vally lint + evals and uploads results.

Review details

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

• Files reviewed: 7/7 changed files
• Comments generated: 4
• Review effort level: Low

[heaths]

@copilot tye eval prompts should be vague. Check that the skills are invoked and the assertions pass, but don't include implementation details of the skills in the prompts. That's not realistic.

[Copilot]

@copilot tye eval prompts should be vague. Check that the skills are invoked and the assertions pass, but don't include implementation details of the skills in the prompts. That's not realistic.

Updated in 3a2b9d9. The prompts are now vague and task-oriented, while expect_skills and the graders still verify the intended skill behavior.

[heaths]

@copilot resolve the active comments using the suggestions given.

[Copilot]

@copilot resolve the active comments using the suggestions given.

Updated in a5127a4.

[heaths]

@copilot resolve the check error "Error: Unable to resolve action actions/setup-node@49933ea5288caeca8642e84d8d5f7a3f7c7bd7e8, unable to find version 49933ea5288caeca8642e84d8d5f7a3f7c7bd7e8" and fix the workflow. Pin the GitHub action using pin-GitHub-actions skill in this repo. Already rename the workflow to be just "eval" for both the file name (eval.yml) and the name in the file.

[Copilot]

@copilot resolve the check error "Error: Unable to resolve action actions/setup-node@49933ea5288caeca8642e84d8d5f7a3f7c7bd7e8, unable to find version 49933ea5288caeca8642e84d8d5f7a3f7c7bd7e8" and ...

Updated in 4a6d841.

[heaths]

@copilot running evals fails with " Error: Authorization error, you may need to run /login (Request ID: DC33:24F025:3306A38:3777493:6A3CE6B2)". Fix the eval workflow. According to https://microsoft.github.io/vally/get-started/add-to-ci/, the GitHub token needs to be passed eg:

name: Skill Evalon:
pull_request:
paths:
- "**/SKILL.md"
- "**/eval.yaml"jobs:
eval:
runs-on: ubuntu-lateststeps:
- uses: actions/checkout@v4
- uses: actions/setup-node@v4with:
node-version: "22"
- run: npm install -g @microsoft/vally-cli
- name: Run evalsrun: | vally eval \ --eval-spec eval.yaml \ --skill-dir . \ --output-dir ./resultsenv:
GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
- name: Upload trajectoriesif: always()uses: actions/upload-artifact@v4with:
name: eval-resultspath: ./results/

[Copilot]

@copilot running evals fails with " Error: Authorization error, you may need to run /login (Request ID: DC33:24F025:3306A38:3777493:6A3CE6B2)". Fix the eval workflow. According to https://microsoft.g...

Updated in e9cfcc2.

[heaths]

@copilot running evals is still failing with "Error: Authorization error, you may need to run /login (Request ID: 2818:77508:338EB17:38288BC:6A3CE88E)". Fix the workflow. If you cannot determine why, do not use trial and error to fix it. Terminate with an error and I'll ask colleagues who might know why.

[heaths]

@chlowell I'm getting an auth error trying to run evals. Any idea why?

Error: Authorization error, you may need to run /login (Request ID: 2818:77508:338EB17:38288BC:6A3CE88E)

I followed the docs and can't find anything about token requirements.

/cc @LarryOsterman