Add output_format A/B experiment to daily-code-metrics workflow

[Copilot]

This updates daily-code-metrics to run an A/B experiment on report verbosity (full_detail vs executive_summary) so engagement, token cost, runtime, and chart volume can be measured from real runs. The workflow prompt now supports variant-specific chart generation and report structure while preserving the existing full-detail behavior as control.

• Experiment wiring in workflow frontmatter

  • Added experiments.output_format with variants, hypothesis, primary/secondary metrics, guardrails, weights, and sampling window.
  • Included a compile-valid numeric issue placeholder required by current schema.

• Variant-gated chart instructions

  • Wrapped chart requirements in conditional template blocks.
  • full_detail: existing 6-chart specification.
  • executive_summary: reduced to 2 charts (quality_score_breakdown, historical_trends).

• Variant-gated report template

  • Added conditional report body branch for executive_summary:
    • key metrics line
    • 2 chart embeds
    • 3 recommendation bullets
    • no <details> metric tables
  • Kept existing full report template in else branch for control.

• Prompt guidance alignment

  • Updated chart/report guidance text to explicitly describe variant-dependent output expectations.
  • Added note to read GH_AW_EXPERIMENTS_OUTPUT_FORMAT in Python logic for branch selection.

• Compiled workflow output

  • Regenerated .github/workflows/daily-code-metrics.lock.yml to include experiment activation/state plumbing.

experiments:
  output_format:
    variants: [full_detail, executive_summary]
    metric: discussion_engagement_score
    secondary_metrics: [output_token_count, run_duration_seconds, chart_count]

{{#if experiments.output_format == 'executive_summary' }}
### 📊 Key Visualizations
...2 charts...
{{else}}
### 📊 Visualizations
...existing full-detail sections...
{{/if}}

[Copilot]

Pull request overview

Adds an output_format A/B experiment to the daily-code-metrics workflow so real-run metrics (engagement, tokens, runtime, chart volume) can be compared between full_detail and executive_summary report variants.

Changes:

• Adds experiments.output_format frontmatter metadata and gates chart + report templates on the selected variant.
• Updates the compiled workflow (.lock.yml) to restore/pick/persist experiment state and plumb GH_AW_EXPERIMENTS_OUTPUT_FORMAT.
• Regenerates daily-model-inventory.lock.yml (appears unrelated to the PR’s stated scope).

Show a summary per file

File	Description
.github/workflows/daily-code-metrics.md	Defines the experiment and adds variant-gated instructions for chart generation and report structure.
.github/workflows/daily-code-metrics.lock.yml	Compiled workflow updates to select variants, pass env vars, and persist experiment state.
.github/workflows/daily-model-inventory.lock.yml	Lock file regen (heredoc marker/hash churn; likely incidental).

Copilot's findings

Tip

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

• Files reviewed: 3/3 changed files
• Comments generated: 5