Avoid duplicate scaffold eval execution

[woud420]

Primary changes

The tiered entrypoint and individual examples can require repeated unchanged runs with unclear completion. Make individual commands iteration alternatives, reuse passing tier results, and require reruns after relevant fixes or nondeterminism evidence.

Before:

fix the template, rerun, and stop iterating only after consecutive clean runs.

After (source excerpt):

After a relevant fix, rerun the affected checks and the required completion tier.

Reviewer walkthrough

• .agents/skills/scaffold-evals/SKILL.md: review the scoped instruction change.
• docs/evals.md: review the scoped instruction change.

Correctness and invariants

The change is limited to audit recommendation 40. Existing authorization and runtime correctness boundaries are preserved except for the explicitly proposed instruction clarification shown in the diff. No installed skill copies, live services, deployment configuration, or unrelated user files are changed.

Testing and QA

• git diff --check — passed
• Inspected the before/after instruction diff and the unchanged source/runtime scope.
• Confirmed full-supported-matrix validation and the release full-tier requirement remain intact.
• Skill-review diagnostics match audit-base exactly; no new errors or warnings. Confirmed full supported matrix for full validation and full-tier release gates remain required; no eval code or CI configuration changed.

Risk / Rollout

Draft for review. Instruction behavior changes only after this source change is accepted and used; no deployment or installed-skill sync is part of this PR. Other audit PRs may touch the same instruction file and should be rebased and reviewed when earlier changes land.

Refs ENG-462

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96.21%. Comparing base (85150d8) to head (63fcb82).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #112   +/-   ##
=======================================
  Coverage   96.21%   96.21%           
=======================================
  Files          58       58           
  Lines        4169     4169           
=======================================
  Hits         4011     4011           
  Misses        158      158           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.