Enforce empirical checkout mandate in PR review protocol

Context During a PR review cycle, the agent approved a PR with a 100/100 [EXECUTION_QUALITY] score ("Tests green") based purely on reading the diff via get_pull_request_diff, without actually running the test suite. This directly violates the Verify-Before-Assert mandate.

The Problem The PR review workflow guide (.agents/skills/pr-review/references/pr-review-guide.md) funnels agents into a "static diff hallucination" trap. Step 1 of the operational mandates explicitly instructs the agent to retrieve state using get_pull_request_diff and get_conversation, but omits any structural requirement to use checkout_pull_request and execute tests prior to generating the review.

The Architectural Reality The Verify-Before-Assert pre-flight check exists (AGENTS.md §2.3) but lacks localized enforcement inside the pr-review protocol's scoring rubric.

The Fix Update .agents/skills/pr-review/references/pr-review-guide.md to include a dedicated "Empirical Checkout Mandate". This step must explicitly require checking out the PR branch locally and executing the test suite to validate execution quality, turning this friction into gold.

Acceptance Criteria

• Insert "Empirical Checkout Mandate" into Section 2 of pr-review-guide.md.
• Explicitly warn that skipping local test execution while claiming "tests pass" is a catastrophic Verify-Before-Assert violation.

Origin Session ID: df8049d8-56ad-417b-ae0e-17a38e22a0ae