Claudetest
Let your agent author Playwright and Appium tests, debug failures, and open GitHub fix PRs from the IDE.
Overview
ClaudeTest MCP is an MCP server for the Ship phase that generates Playwright and Appium tests, diagnoses failures, and opens GitHub fix PRs from your agent.
What is this MCP server?
- Playwright and Appium test automation exposed to Claude via stdio MCP (@fasteningcode/claudetest v0.0.3)
- Generates tests, diagnoses failures, and can open fix pull requests on GitHub
- Requires GITHUB_TOKEN with repo and workflow scopes
- Targets web and mobile E2E workflows from the agent instead of context-switching to CI logs alone
- npm registryType package with stdio transport
- Published version 0.0.3 on npm (@fasteningcode/claudetest)
- Supports Playwright (web) and Appium (mobile) per server description
- One required secret: GITHUB_TOKEN
What problem does it solve?
You know E2E tests matter, but writing Playwright or Appium specs and chasing red CI jobs eats the hours you wanted for features.
Who is it for?
Indie devs on GitHub who want agent-driven Playwright/Appium coverage and PR-based fixes from Claude or Cursor.
Skip if: Repos without GitHub, teams that forbid PAT-based PR automation, or projects that only need static analysis with no E2E layer.
What do I get? / Deliverables
Your agent can scaffold tests, explain failures, and land fix PRs so shipping stays guarded without a dedicated QA hire.
- Generated or updated Playwright/Appium test files aligned to your flows
- Failure analysis narratives and optional GitHub pull requests with fixes
Recommended MCP Servers
Journey fit
Ship is the first moment automated tests block bad releases, so testing is the canonical shelf for this MCP server. Testing is where generated Playwright/Appium suites, failure diagnosis, and PR-based fixes directly reduce launch risk.
How it compares
Agent-facing test generator and failure copilot, not a JSON-only Chrome pool runner like dedicated E2E MCP servers.
Common Questions / FAQ
Who is ClaudeTest MCP for?
Solo builders and small teams using Claude with GitHub who want Playwright or Appium tests generated and maintained inside the agent workflow.
When should I use ClaudeTest MCP?
Use it before release or after CI failures when you need new E2E specs, root-cause hints, or an automated fix PR—not for one-off URL screenshots.
How do I add ClaudeTest MCP to my agent?
Add the npm package @fasteningcode/claudetest via stdio, set GITHUB_TOKEN with repo and workflow scopes, then enable the server in your MCP client config.