Task: Set Up Skill Review Pipeline in Azure DevOps

Our organization uses Azure DevOps for source control and CI/CD. We have an internal repository at ~/project hosted in Azure Repos. We want automated skill review on PRs and pushes.

Configuration (use these values directly — do not ask configuration questions)

• CI Platform: Azure DevOps
• Architecture: Single-pipeline (all internal team members)
• Target branch: release
• Trigger paths: **/SKILL.md, **/skills/** (defaults)
• Cache location: ci/review-cache/scores.json (custom — not the default)
• Score threshold: None (informational only — no quality gate)

What the pipeline should do

1. Run on PRs targeting release and on pushes to release
2. Detect which skill files changed
3. Run quality reviews on each changed skill
4. Calculate and display scores with comparison to previous runs
5. Post review results as a PR comment thread using the Azure DevOps REST API (not GitHub API) with System.AccessToken
6. Scores are informational only — no pass/fail gating
7. Cache scores and auto-commit the cache to release after merges

Current repository structure

~/project/
  .git/
  skills/
    data-pipeline/SKILL.md
    ml-ops/SKILL.md
  ci/
    build.yml

Expected output files

1. ~/project/azure-pipelines.yml — Azure DevOps pipeline YAML
2. ~/project/ci/review-cache/scores.json — initialized cache file at custom location

Write all output files to the ~/project directory. Do not run or test the pipeline.