feat(calibration): size-constraints strip responsive visual-compare (…

[let-sunny]

…#205)

• converter.md: mandatory expanded-viewport compare for size-constraints strip; baselineResponsiveSimilarity from step 6 responsiveSimilarity
• evaluation: apply strip override for responsive-critical only when size-constraints.responsiveDelta is finite
• critic.md: document conditional strip vs baseline responsive
• tests: strip ablation applies when strip responsive delta present

Made-with: Cursor

Summary by CodeRabbit

• Bug Fixes

  • More accurate responsive similarity measurement at expanded viewport; size-constraint comparisons are now required and produce explicit responsive metrics when available.
  • Responsive-critical rule handling refined so strip-based overrides apply only when a valid size-constraint responsive metric exists; otherwise baseline responsive classification is preserved.

• Tests

  • Added and updated tests to verify responsive-critical behavior across finite and non-finite metric scenarios and to ensure correct inclusion/exclusion of strip-based overrides.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 29d8f62d-e2a7-464a-afe0-e78723b7d15d

📥 Commits

Reviewing files that changed from the base of the PR and between ab232cc and fcaa5e7.

📒 Files selected for processing (1)

• src/agents/evaluation-agent.test.ts

---

📝 Walkthrough

Walkthrough

The pull request requires responsive comparison at the expanded viewport for size-constraints strips and conditions responsive-critical rule overrides on the presence of a finite size-constraints.responsiveDelta, while keeping responsive fields null for other strip types when step 6 is skipped.

Changes

Cohort / File(s)	Summary
Converter & Critic docs .claude/agents/calibration/converter.md, .claude/agents/calibration/critic.md	Clarify responsive similarity instructions: always measure responsive fields for size-constraints at expanded viewport (run visual-compare and record responsiveSimilarity/responsiveDelta); make responsive-critical override behavior conditional on a finite size-constraints.responsiveDelta.
Evaluation agent implementation src/agents/evaluation-agent.ts	Add sizeConstraintsStripHasResponsiveMetric() and change strip-override logic to skip responsive-critical overrides only when size-constraints.responsiveDelta is missing or non-finite; allow strip-based classification when a finite responsiveDelta exists.
Tests src/agents/evaluation-agent.test.ts	Update test descriptions; add cases asserting responsive-critical overrides are applied when size-constraints.responsiveDelta is finite and not applied when it is non-finite (NaN, +Infinity, -Infinity).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~22 minutes

Possibly related issues

• feat: add size-constraints to calibration strip ablation #204: Related — touches size-constraints integration and strip-ablation/calibration flow.
• feat: size-constraints strip comparison at responsive viewport #205: Related — focuses on running size-constraints comparison at expanded/responsive viewport and recording responsive deltas.

Possibly related PRs

• feat: integrate ablation into calibration pipeline for objective difficulty #194: Related — modifies converter/critic/evaluation-agent logic around strip-ablation and responsive deltas.
• feat: responsive viewport comparison in calibration (#182) #183: Related — refines responsive-comparison and responsiveDelta handling used by evaluation overrides.
• feat: strip ablation enhancements for accurate calibration #208: Related — earlier changes to evaluation-agent handling of responsive-critical strips that this PR tightens with metric gating.

Poem

🐰 In widened view the strips align,

size-constraints sings its responsive sign.
When delta is finite, ablation may play,
otherwise the baseline holds the day.
A rabbit cheers for metrics kept in mind.

🚥 Pre-merge checks | ✅ 2

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly summarizes the main change: implementing mandatory responsive visual-compare for size-constraints strip at expanded viewport.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/205-size-constraints-responsive-strip

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@src/agents/evaluation-agent.test.ts`:
- Around line 536-573: Add a test that explicitly passes a non-finite
responsiveDelta (e.g., NaN or Infinity) to the runEvaluationAgent call to
exercise the production gate that treats non-finite like missing values;
duplicate the existing test case structure using EvaluationAgentInput, set
responsiveDelta to NaN (or Infinity), include the same stripDeltas and rule data
for "missing-size-constraint", then assert the resulting mismatch for ruleId
"missing-size-constraint" has the same expected actualDifficulty and contains
"strip-ablation" (or whatever the current null/missing behavior asserts). Ensure
the new test references runEvaluationAgent, EvaluationAgentInput,
responsiveDelta, stripDeltas, and the ruleId "missing-size-constraint" so it's
clear which logic is being validated.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro

Run ID: 44208f00-d149-4027-b560-c5a16343ff54

📥 Commits

Reviewing files that changed from the base of the PR and between fb45b54 and ab232cc.

📒 Files selected for processing (4)

• .claude/agents/calibration/converter.md
• .claude/agents/calibration/critic.md
• src/agents/evaluation-agent.test.ts
• src/agents/evaluation-agent.ts