karote00/asyra/unit-failure-visual-replay
Claude Code and agent skill by karote00/asyra
Trust tier 3
- community
Source check pending
Checking that the SKILL.md file is still in the repository.
Use when unit or integration tests emit source-space failure manifests and the user needs browser replay screenshots with markers showing where each failure appears in the app. Applies to canonical stroke tests and future geometry/UI tests that can map failures to fixture-local coordinates.
Install
npx vskill@latest install karote00/asyra/unit-failure-visual-replaySKILL.md
Unit Failure Visual Replay
Trigger Signals
Use this skill when:
- A unit/integration test fails and can emit a source-space failure manifest.
- The user asks to see where a unit failure appears on the actual app canvas.
- A test failure contains geometry coordinates, source segment IDs, anchor IDs, terminal IDs, or fixture-local points.
- A visual replay should be generated without making the unit test depend on browser/E2E runtime.
Do Not Use When
- The failure cannot be mapped to fixture-local coordinates or a stable fixture.
- The user only needs the raw test log.
- The requested validation is already a normal E2E screenshot without a unit manifest.
- The replay would require operating outside the project artifact directories.
Required Inputs
- Unit/integration test command that emits a manifest.
- Manifest path.
- E2E replay spec path.
- Existing app URL resolved from
APP_URL. - Fixture reconstruction rules keyed by
caseKeyandfixtureKind.
For canonical stroke matrix:
- Unit command:
- Replay command:
Preflight
- Resolve
APP_URLand confirm the relevant app server is already available there. - Confirm the unit test writes or clears the manifest every run.
- Confirm the replay spec reads JSON only, not unit test source.
- Confirm artifact paths are inside the project.
- Confirm stale manifest behavior: passing unit tests must remove or overwrite old failure data.
Deterministic Procedure
- Run the unit/integration test.
- A non-zero exit code is expected when product bugs are present.
- Do not weaken assertions or delete the manifest to make the command pass.
- Inspect the manifest.
- Read
failureCount. - Confirm
failures.length === failureCount. - Confirm every failure has
markerId,errorCode,caseKey,fixtureKind,localPoint, andrecommendedViewport.
- Run the replay E2E spec.
- Rebuild the fixture from
caseKeyandfixtureKind. - Focus the viewport using
recommendedViewport.zoomandrecommendedViewport.center. - Draw the marker at
localPoint. - Screenshot each marker.
- Write a report linking marker IDs to screenshots.
- Check replay outputs.
- Screenshot count should match manifest
failureCount, unless the replay intentionally caps samples and documents the cap. - Marker labels must be visible.
- Report rows must include marker ID, error code, case key, source location, summary, and screenshot link.
Validation Matrix
Validate these conditions:
- Manifest generated from the current unit run, not stale.
- Replay E2E passes.
- Number of marker screenshots matches manifest failure count.
- Report exists and links to screenshots.
- At least one representative screenshot per
errorCodeis visually inspectable. - Targeted lint passes for edited test/replay files.
Required Output Format
Report concisely:
- Unit command result and failed test count.
- Manifest failure count.
- Replay E2E result.
- Marker screenshot count.
- Manifest/report paths.
- Representative screenshots when the user asks to see them.
Guardrails
- Unit tests must remain browser-independent.
- E2E replay must not import unit test files.
- Do not use screenshot generation as a substitute for unit assertions.
- Do not start an extra dev server if
APP_URLis already serving the app. - Do not store artifacts outside the project.
- Keep repetitive failures capped deterministically so replay remains usable.
Failure Policy
- If manifest is missing after a failing unit test, fix the manifest emission first.
- If replay cannot reconstruct a fixture, add explicit
fixtureKind/caseKeyhandling. - If marker placement is wrong, fix
localPointor fixture rect mapping before trusting screenshots. - If E2E replay fails because the app is unavailable, report that environment issue separately from product failures.
- If the unit test passes, stale failure artifacts should not be treated as current evidence.
Manifest Contract
Required per failure:
markerId: stable marker likeF001.errorCode: compact taxonomy such asSTROKE_OVERLAP,CAP_EXTENSION_MISSING,CAP_SHAPE_MISMATCH,SEGMENT_ENVELOPE_MISMATCH,JOIN_SIGNATURE_MISSING, orPIPELINE_INCOMPLETE.caseKey: enough to rebuild fixture and stroke state.summary: one-line diagnosis.fixtureKind: fixture builder selector, such asself-check-staroropen-line.localPoint: fixture-local marker point.recommendedViewport:{ zoom, center }in fixture-local coordinates.
Recommended:
sourceSegmentId,sourcePointId,nearestAnchorId.talong source segment.side:inside,outside,center,terminal,join, oroverlap.expectedandactualcompact values.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingdocs/ai/skills/unit-failure-visual-replay/SKILL.md
Popularity
- Repository stars
- 44
- Repository forks
- 2
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/karote00/asyra/unit-failure-visual-replay)