Add tone_variant A/B campaign to breaking-change-checker workflow

[Copilot]

This updates breaking-change-checker to run an A/B experiment on reporting tone (neutral vs urgent) without changing breaking-change detection logic. The campaign metadata and prompt conditionals were added so tone can be assigned at runtime and compared against engagement and guardrail metrics.

• Experiment frontmatter

  • Added experiments.tone_variant in rich-object form with:
    • variants, description, hypothesis
    • metric, secondary_metrics
    • guardrail_metrics (run_success_rate, empty_output_rate)
    • min_samples, weight, start_date, issue

• Prompt-body tone branching

  • Added variant-aware handlebars conditionals in the issue-generation instructions:
    • tone guidance block (urgent vs neutral)
    • Summary -> Status line (Requires Immediate Review vs Review Recommended)
    • Recommendations wording (prioritized risk/action framing vs neutral guidance)

• Compiled workflow output

  • Regenerated breaking-change-checker.lock.yml from the markdown workflow so runtime experiment selection and outputs reflect the new campaign.

{{#if experiments.tone_variant == "urgent" }}
- **Status**: ⚠️ Requires Immediate Review
{{else}}
- **Status**: Review Recommended
{{/if}}

[pelikhan]

@copilot check conditional syntax

[Copilot]

Pull request overview

Adds an A/B experiment (experiments.tone_variant) to the breaking-change-checker workflow so the issue-reporting prompt can branch between neutral and urgent tone variants while keeping breaking-change detection logic unchanged.

Changes:

• Introduces experiments.tone_variant frontmatter metadata (variants, hypothesis, metrics, guardrails, sampling/weighting, tracking issue).
• Adds variant-aware handlebars conditionals to adjust tone guidance and issue template fields (Status + Recommendations wording).
• Regenerates breaking-change-checker.lock.yml to include runtime experiment selection, persistence, and prompt substitutions.

Show a summary per file

File	Description
.github/workflows/pr-code-quality-reviewer.lock.yml	Unrelated gh-aw metadata hash churn included in this PR (does not support the stated PR purpose).
.github/workflows/breaking-change-checker.md	Adds the tone_variant experiment configuration and prompt-body tone branching.
.github/workflows/breaking-change-checker.lock.yml	Updates compiled workflow to pick/restore/persist experiment state and pass tone variant into prompt substitution.

Review details

Tip

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

• Files reviewed: 3/3 changed files
• Comments generated: 1
• Review effort level: Low

[github-actions]

🤖 PR Triage — §28486872548

Field	Value
Category	feature
Risk	medium
Score	28 / 100
Priority	low
Action	⏸ defer

Score breakdown: Impact 15/50 · Urgency 5/30 · Quality 8/20

Batch: draft-wip (PRs #42595, #42596, #42597)

Rationale: WIP/Draft A/B tone experiment for breaking-change checker. Low urgency, no reviews yet. Should be completed and marked ready before further review.

Labels applied: pr-type:feature pr-risk:medium pr-priority:low pr-action:defer pr-batch:draft-wip pr-agent:copilot-swe-agent

Generated by 🔧 PR Triage Agent · 77.2 AIC · ⌖ 9.82 AIC · ⊞ 1.6K · ◷

[Copilot]

@copilot check conditional syntax

Fixed in df78190.

[github-actions]

Hey @app/copilot-swe-agent 👋 — nice work setting up the tone_variant A/B campaign! The experiment frontmatter is well-structured (hypothesis, guardrail metrics, min_samples, 50/50 weight) and the handlebars branching in the prompt body is clean and readable.

A couple of things to tighten up before this lands:

• Add tests for the tone branching — the new {{#if experiments.tone_variant == 'urgent'}} conditionals in breaking-change-checker.md aren't covered by any test changes. At minimum, consider adding snapshot or render tests that exercise both the urgent and neutral paths and assert the correct Status line and Recommendations wording appears in each case.
• Explain the pr-code-quality-reviewer.lock.yml body-hash change — the diff touches that file (only the body_hash in the metadata comment changes; no corresponding .md source file is modified). This looks like it may be a stale or incidental recompilation artifact. If expected, a one-liner in the PR description noting why would help reviewers confirm it's benign.

If you'd like a hand, you can assign this prompt to your coding agent:

Add tests for the tone_variant A/B experiment in .github/workflows/breaking-change-checker.md.

Cover the following scenarios:
1. When experiments.tone_variant == 'urgent': assert the rendered output contains "Requires Immediate Review" in the Status line and the urgent Recommendations copy.
2. When experiments.tone_variant == 'neutral' (else branch): assert the rendered output contains "Review Recommended" in the Status line and the neutral Recommendations copy.
3. Verify both branches emit the shared structural fields (Title, Date, Total Breaking Changes, Severity, Commits Analyzed) unchanged.

Also investigate whether the body_hash change in .github/workflows/pr-code-quality-reviewer.lock.yml is an expected compilation side-effect or an accidental inclusion, and either remove it or add a note in the PR description explaining it.

Generated by ✅ Contribution Check · 360.2 AIC · ⌖ 15.7 AIC · ⊞ 6.3K · ◷

[github-actions]

🎉 This pull request is included in a new release.

Release: v0.82.2