feat(judge): retain per-platform/per-layer detail in JudgeResult#73
Merged
Conversation
Step 1 of the HTML validation report (docs/validation-report.md). runJudge already computed full Layer 1 findings and Layer 2 command/ duration/stderr inside evaluate(), then discarded everything but counts and a summary string. Widen evaluate() to return the full Layer1Result + Layer2Result, and surface them on a new optional, additive JudgeResult.platforms (readonly PlatformDetail[]). Each PlatformDetail carries layer1 (pass + findings), layer2 (pass, command, mode, exitCode, durationMs, stderrTail), and the matching layer3 VisualJudgePlatformReport for ios/android (rails has none). Back-compatible: existing consumers (CLI summary, MCP) ignore the new field; the stub judge path omits it. This is the structured data the validation report renderer will consume in a follow-up. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
dadachi
added a commit
that referenced
this pull request
May 22, 2026
#74) Implements steps 3-7 of docs/validation-report.md (step 1-2 shipped in #73). Every run now writes out/<slug>/report.json (machine-readable RunReport) and validation-report.html (self-contained, screenshots base64-embedded) summarizing all validation layers + the reviewer. - src/report/model.ts — RunReport / RunMeta / RepairAttempt / AssetMap - src/report/render.ts — renderReport(report, assets): pure, no I/O; gate strip, platform×layer matrix, Layer 1 findings, Layer 2 stderr, Layer 3 screenshots + rubric rationales (incl. Stage 2 filmstrip), reviewer diff, rename-plan appendix, reproduce footer. HTML-escaped throughout. - src/report/theme.ts — inline brand CSS (matches social-preview) - src/report/collect.ts — buildRunReport (pure assembly), resolveAssets (embed as data: URI, or copy to report-assets/), writeReport (report.json + HTML) - src/version.ts — shared readPackageVersion (dedups mcp.ts copy) - dispatch: records timing, assembles + writes the report, returns DispatchResult = JudgeResult & { report, reportPaths }. Writing defaults off in stub mode so the test suite never touches ./out. - index: --no-report / --report-format / --report-embed / --report-open flags; prints a file:// path. mcp: returns report + html/json paths in structuredContent. Tests: pure renderer assertions (findings, stderr, diff, rename plan, HTML escaping, empty-platforms summary fallback), writeReport round-trip (self-contained — asserts no tmp/ paths and no external URLs leak), embed=false externalization, and buildRunReport assembly. 46/46 pass. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
dadachi
added a commit
that referenced
this pull request
May 22, 2026
…kflow (#76) Adds section 10 to docs/validation-report.md covering how to render and visually verify validation-report.html: - Path A — Playwright MCP (interactive, inside Claude Code): local-scope `claude mcp add playwright`, restart to load tools, then browser_navigate (file:// URL) + browser_take_screenshot. Note it's a maintainer convenience, not part of the pipeline (device shots use mobile-mcp). - Path B — Playwright CLI (headless, no MCP, CI-friendly): `npx -y playwright@latest screenshot --full-page file://… report.png`, no project dependency added. Also marks section 9 as shipped (#73/#74/#75). Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Step 1 of the HTML validation report (
docs/validation-report.md).runJudgealready computed full Layer 1 findings and Layer 2 command/duration/stderrTailinsideevaluate(), then discarded everything but counts and a summary string — so the report data never reached the caller. This PR widensevaluate()to return the fullLayer1Result+Layer2Resultand surfaces them on a new field.Changes
JudgeResultgains an optional, additiveplatforms?: readonly PlatformDetail[].PlatformDetailtype:layer1(pass + findings),layer2(pass, command, mode, exitCode, durationMs, stderrTail), and the matchinglayer3VisualJudgePlatformReportfor ios/android (rails has none).evaluate()returns the full layer results; trace lines + pass aggregates updated to the new shape (no behavior change to logs or summary).Compatibility
generate_app) ignore the new field.platforms(unchanged).npm run cigreen locally — build + typecheck + 41/41 tests pass.Renderer +
report.jsonemission come in follow-up PRs (steps 3–7 in the spec).🤖 Generated with Claude Code