# Openwork Chrome MCP Testing

> Use Chrome MCP to verify OpenWork UI flows, especially any feature that touches remote behavior. Triggers when user mentions: - "test with chrome mcp" - "ui verification" - "remote behavior test"

- Skill: `different-ai/openwork-chrome-mcp-testing` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add different-ai/openwork-chrome-mcp-testing`
- Raw SKILL.md: https://api.skillmd.com/api/skills/different-ai/openwork-chrome-mcp-testing/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: different-ai (https://skillmd.com/u/different-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/different-ai/openwork-chrome-mcp-testing

---


## Quick Usage (Already Configured)

### 1) Start headless + web UI

Start the OpenWork dev stack via Docker and open the UI:
- Use `skills/openwork-docker-chrome-mcp/SKILL.md` step 1.
- Confirm the Web UI loads and a session/chat view is accessible.

### 2) Test the feature in the UI

- Navigate to the relevant screen.
- Perform the feature flow end-to-end.
- If the change touches remote behavior, verify the remote effect is visible in the UI.

### 3) Capture evidence

- Take a Chrome MCP snapshot.
- Pull console logs with Chrome MCP.
- Attach logs if something fails:
  - capture Chrome console output via Chrome MCP
  - optionally capture Docker logs for the dev stack

## Required Gate (Non-Negotiable)

- Any feature that changes remote behavior MUST be validated through Chrome MCP.
- The feature is NOT done until the UI action succeeds in the browser via Chrome MCP.
- If Chrome MCP validation is skipped, the change is considered incomplete and must not be merged.

## Common Gotchas

- If OpenCode Engine stays disconnected, ensure the OpenWork URL is clean and the client token is correct.
- If you see CORS errors, check OpenWork server CORS headers for custom OpenCode headers.