stevekaplanai/agenthost/visual-gate
Claude Code and agent skill by stevekaplanai/agenthost
- community
Checking that the SKILL.md file is still in the repository.
This skill should be used when the user asks to "run the visual gate", "capture the evidence", "screenshot this change", "/visual-gate", before merging any change to a visible surface, or when a completion claim is about to be made without rendered proof.
Install
npx vskill@latest install stevekaplanai/agenthost/visual-gateSKILL.md
Visual Gate — Produce the Stage Evidence
Capture the visual evidence the current change requires — screenshots at both required viewports, before/after pairs — and emit a manifest a reviewer or a reviewer agent can verify. This is the Build/Change stage of the visual-first contract (see the visual-first-design-skill).
Steps
- Determine what changed and where it renders. From the argument, or by reading the diff (
git diff main...HEAD --name-only) and mapping changed files to the pages/routes they affect. If a changed file renders nowhere reachable, that is a finding — report it; there is nothing to screenshot because the work is not wired in.
- Capture BEFORE. Render the unchanged state — the main branch, the deployed URL, or the pre-change commit — at 390px and 1280px:
``bash mkdir -p visual-evidence/<change-slug> npx playwright screenshot --viewport-size=390,844 --full-page "$BEFORE_URL" "visual-evidence/<change-slug>/before-390.png" npx playwright screenshot --viewport-size=1280,800 --full-page "$BEFORE_URL" "visual-evidence/<change-slug>/before-1280.png" ``
If the change is brand new (nothing existed before), capture the surrounding page as-is — the pair still shows what the change did to it.
- Capture AFTER. Same two viewports, same routes, same scroll position, from the running changed code (dev server or preview deploy). Same filenames with
after-.
- Verify every file exists.
ls -la visual-evidence/<change-slug>/— four PNGs minimum. A capture command that failed silently produces a manifest entry that is a lie; check before writing the manifest.
- Write the manifest.
visual-evidence/<change-slug>/manifest.md: a table of artifact → absolute path → viewport → source URL → capture timestamp, plus one line naming anything that could NOT be captured and why. The manifest never lists a file that is not on disk.
- Deliver. The before/after pairs (phone pair first), the manifest path, and a one-line plain-language statement of what the pair shows. This evidence rides with the PR — commit
visual-evidence/on the branch so the reviewer agent and the human can reach it.
Rules
- 390px and 1280px are the floor, not the ceiling. Add viewports only when the change targets one.
- Before and after must be same-route, same-viewport, same-scroll — otherwise the pair proves nothing.
- The dev server proves the code; only the deployed URL proves the deploy. After a deploy, re-run the AFTER captures against the live URL and add them as
live-390.png/live-1280.png. - Never write a manifest entry for an artifact that does not exist on disk. An honest "could not capture X because Y" is a valid manifest line; a fabricated entry is the worst possible output.
Security
- Security scanPassed, 93 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingcontainer/skills-preload/visual-gate/SKILL.md
Popularity
- Repository stars
- 2
- Repository forks
- 1
- 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/stevekaplanai/agenthost/visual-gate)