# Visual Evidence

> Use for UI changes in an agent-md repository; captures a screenshot and writes the structured markdown evidence required by the visual hook.

- Skill: `iamfakeguru/visual-evidence` (Agent Skill)
- Install (CLI): `npx skillmds@latest add iamfakeguru/visual-evidence`
- Raw SKILL.md: https://api.skillmd.com/api/skills/iamfakeguru/visual-evidence/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: iamfakeguru (https://skillmd.com/u/iamfakeguru)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/iamfakeguru/visual-evidence

---


# Visual Evidence

Use this skill for any UI-facing change.

1. Build or start the app in the normal project-specific way.
2. Capture the changed route with:

```bash
./.agent-md/bin/playwright-capture.sh <url> .agent/visual/<name>.png
```

3. Write a markdown note next to the image under `.agent/visual/`.
4. The note must reference the image filename and include these fields:

```markdown
# Visual Check

Changed files:
- <path>

Route: <url or route>
Viewport: <width>x<height>
Artifact: <image filename>
Observed result: <what was verified>
```

5. Use an independent verifier when possible: another agent, a browser screenshot inspection, or the human.
6. Do not self-grade visual correctness from code alone.