optimize: reduce token usage in three high-consumption workflows

[Copilot]

✨ Enhancement

Three workflows were identified (via run IDs 25686395186, 25686664431, 25686664489) as disproportionately high token consumers due to uncapped continuation loops and live API calls inside the agent loop.

What does this improve?

• test-quality-sentinel — max-continuations: 40 → 15 (~63% worst-case reduction)
• mattpocock-skills-reviewer — adds max-continuations: 10; PR diff and metadata are now pre-fetched into /tmp/gh-aw/agent/ before the agent starts, eliminating redundant live gh pr diff/gh pr view calls inside the loop
• contribution-check — adds max-continuations: 20 as a hard engine-level cap, complementing the existing 50-turn soft limit in the prompt to guard against subagent retry runaway

Implementation approach:

• mattpocock-skills-reviewer: new Pre-fetch PR diff pre-agent step writes pr-diff.patch + pr-meta.json to disk; Step 1 of the agent prompt updated to read these files instead of making API calls
• All three engine configs now carry explicit max-continuations ceilings
• .lock.yml files auto-regenerated; all three workflows compile with 0 errors

[Copilot]

Pull request overview

Reduces worst-case token consumption in three high-usage agentic workflows by capping continuation loops and removing redundant live GitHub API calls from inside an agent loop.

Changes:

• Add/adjust engine.max-continuations ceilings for test-quality-sentinel, mattpocock-skills-reviewer, and contribution-check.
• Add a pre-agent-steps PR prefetch step to mattpocock-skills-reviewer that writes PR metadata + full diff to /tmp/gh-aw/agent/, and update the prompt to read those files instead of calling gh pr view/diff.
• Regenerate .lock.yml files and update action pin data to include github/gh-aw-actions/setup@v0.71.5.

Show a summary per file

File	Description
pkg/workflow/data/action_pins.json	Adds pin entry for github/gh-aw-actions/setup@v0.71.5 to support compiled workflow actions.
pkg/actionpins/data/action_pins.json	Mirrors the same new action pin in the actionpins package dataset.
.github/workflows/test-quality-sentinel.md	Lowers Copilot continuation ceiling to reduce worst-case runaway cost.
.github/workflows/test-quality-sentinel.lock.yml	Regenerated compilation output reflecting new continuation cap (--max-autopilot-continues 15).
.github/workflows/mattpocock-skills-reviewer.md	Adds max-continuations: 10 and prefetches PR diff/meta before the agent; prompt updated to read local files.
.github/workflows/mattpocock-skills-reviewer.lock.yml	Regenerated compilation output including new prefetch step and continuation cap (--max-autopilot-continues 10).
.github/workflows/issue-arborist.lock.yml	Regenerated compilation output (includes Codex harness argument formatting updates).
.github/workflows/contribution-check.md	Adds max-continuations: 20 to cap Copilot continuations at the engine level.
.github/workflows/contribution-check.lock.yml	Regenerated compilation output reflecting the new continuation cap (--max-autopilot-continues 20).

Copilot's findings

Tip

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

• Files reviewed: 9/9 changed files
• Comments generated: 0