Use mai-code-1-flash-picker model in pr-triage-agent

[Copilot]

pr-triage-agent ran on the default model tier. Pinning it to mai-code-1-flash-picker lowers recurring per-run AI credit cost for a scheduled (every 6h) workflow.

Changes

• .github/workflows/pr-triage-agent.md — added top-level model: mai-code-1-flash-picker, matching the convention used by other workflows in this repo (issue-arborist.md, terminal-stylist.md, smoke-copilot-mai.md); no workflow uses the nested engine.model form.
• .github/workflows/pr-triage-agent.lock.yml — regenerated via gh aw compile.

model: mai-code-1-flash-picker
engine:
  id: copilot
  copilot-sdk: true

Note for reviewers

The compiler propagates a top-level model: to the detection and evals steps as well, so the lock file changes COPILOT_MODEL: detection → mai-code-1-flash-picker and GH_AW_EVALS_MODEL: "evals" → "mai-code-1-flash-picker". This is existing compiler behavior for any workflow declaring a top-level model (see smoke-copilot-mai.lock.yml), not something introduced here — but worth confirming it's acceptable for the evals judge in this workflow.

• Fixes [q] Use mai-code-1-flash-picker model in pr-triage-agent to reduce cost #55404

[github-actions]

Great work on the cost optimization! 🎯 This PR looks solid:

✅ Focused change — pins the model for pr-triage-agent to reduce recurring AI credit costs, matching the convention used in issue-arborist.md, terminal-stylist.md, and smoke-copilot-mai.md.

✅ Well-explained — the description clearly walks through the motivation, the YAML changes, and importantly flags the compiler behavior propagating the top-level model: to detection and evals steps (with reference to smoke-copilot-mai.lock.yml as precedent).

✅ Ready for review — no additional work needed. The lock file regeneration via gh aw compile and the consistency with existing workflows make this straightforward.

Generated by ✅ Contribution Check · copilot · auto · 56.2 AIC · ⌖ 6.18 AIC · ⊞ 9.3K · ◷

[Copilot]

Pull request overview

Pins the scheduled PR triage workflow to a lower-cost supported model.

Changes:

• Configures mai-code-1-flash-picker.
• Regenerates agent, detection, and eval model settings.

Security regressions

None identified; permissions and guardrails are unchanged.

Validation/scanner results

The model exists in the repository catalog, and lock-file changes match compiler behavior.

Optimization opportunities

Recent runs consumed 265.5 AIC across 10 executions, supporting this optimization.

Show a summary per file

File	Description
.github/workflows/pr-triage-agent.md	Selects the cost-optimized model.
.github/workflows/pr-triage-agent.lock.yml	Applies the model throughout generated jobs.

Review details

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.

• Files reviewed: 2/2 changed files
• Comments generated: 0
• Review effort level: Balanced