Fix Daily Cache Strategy Analyzer codex model variant to avoid model-not-found failures

[Copilot]

Daily Cache Strategy Analyzer was failing consistently in codex runs due to a recurring model-not-found path on the mini variant. This updates the experiment model set to use a codex-available mini variant so scheduled runs can complete and stop emitting the recurring failure signature.

• Workflow experiment model correction

  • Updated .github/workflows/daily-cache-strategy-analyzer.md:
    • experiments.model_size.variants from [gpt-5.4, gpt-5-mini] to [gpt-5.4, gpt-5.4-mini]
    • hypothesis text to match the new mini variant

• Compiled workflow sync

  • Regenerated .github/workflows/daily-cache-strategy-analyzer.lock.yml so runtime experiment spec matches source workflow frontmatter.

• Guardrail test alignment

  • Updated pkg/workflow/prompts_test.go (TestDailyCacheStrategyAnalyzerUsesCodexCompatibleModelsForExperiment) to assert the new expected variant set.

experiments:
  model_size:
    variants: [gpt-5.4, gpt-5.4-mini]

[github-actions]

Hey @Copilot 👋 — good start on tracking down the binary path issue in the Daily Cache Strategy Analyzer! The two investigation steps are already checked off, which is a solid foundation. A couple of things to round this out before it's ready for review:

• Add a prose description — the PR body is currently a task checklist only. A short paragraph explaining what binary path is wrong, why the Daily Cache Strategy Analyzer fails to resolve it, and what the intended fix looks like will help reviewers understand the change without needing to dig through CI logs.
• Add tests — once the fix lands, a test covering the gh-aw binary/model resolution path in the analyzer (happy path + failure case) will prevent regressions.

When you're ready to implement, here's a prompt you can hand off to your coding agent:

Complete the fix for the codex gh-aw binary path in the Daily Cache Strategy Analyzer workflow.

1. Identify where the current binary/model resolution fails — check the workflow config and any runner-side lookup logic.
2. Apply the minimal change to resolve the path correctly (e.g. absolute path, PATH entry, or config key fix).
3. Add a test that validates binary resolution succeeds and fails gracefully when the binary is absent.
4. Update the PR body with: (a) a one-paragraph description of the root cause, (b) what changed, and (c) why this fix is correct.
5. Run the security/secret checks and remove the [WIP] prefix from the PR title when done.

Generated by ✅ Contribution Check · 236.7 AIC · ⌖ 17.2 AIC · ⊞ 6K · ◷

[Copilot]

Pull request overview

This pull request updates the Daily Cache Strategy Analyzer workflow’s experiment model variants to use a codex-available mini variant, preventing recurring model-not-found failures in scheduled codex runs and keeping the compiled workflow in sync.

Changes:

• Updated the workflow frontmatter experiment variants from gpt-5-mini to gpt-5.4-mini and aligned the hypothesis text accordingly.
• Regenerated the compiled .lock.yml so GH_AW_EXPERIMENT_SPEC matches the updated frontmatter.
• Updated the guardrail test to assert the new expected variant set.

Show a summary per file

File	Description
pkg/workflow/prompts_test.go	Adjusts the frontmatter assertion test to expect gpt-5.4-mini instead of gpt-5-mini.
.github/workflows/daily-cache-strategy-analyzer.md	Updates experiment variants and hypothesis to use a codex-compatible mini variant.
.github/workflows/daily-cache-strategy-analyzer.lock.yml	Syncs compiled workflow metadata and embedded experiment spec to the updated frontmatter.

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: 0