BugBug Plugin

BugBug Plugin is the recommended way to connect an AI coding agent to BugBug. It installs BugBug MCP and BugBug Skills together, so the agent has both live BugBug tools and workflow guidance.

BugBug MCP access requires a Pro or higher BugBug plan. BugBug Skills are available on all plans.

First successful setup

Install with the BugBug CLI, then let your AI client authenticate with BugBug. Choose one of the supported clients below. For another Streamable HTTP MCP client, use the manual configuration on the BugBug MCP page.

1

Install for your AI client

Claude Code

npx @bugbug-io/cli plugin --agent=claude

Codex

npx @bugbug-io/cli plugin --agent=codex

Cursor

npx @bugbug-io/cli plugin --agent=cursor

GitHub Copilot

npx @bugbug-io/cli plugin --agent=copilot

VS Code

npx @bugbug-io/cli plugin --agent=vscode

2

Restart and sign in

Restart the AI client if it does not load the plugin immediately, then complete its MCP OAuth sign-in flow.

3

Confirm the connection with a read-only request

Ask the agent for a small, safe action in the BugBug project you expect it to access:

Verify the BugBug connection

Confirm that BugBug is connected. Read my available BugBug projects and list their names only. Do not create, update, run, import, or delete anything. If you cannot authenticate or access a project, tell me exactly what is missing.

If the agent can return the expected projects, you are ready to plan, author, review, and debug BugBug tests.

What it provides

• BugBug MCP with 50 tools for reading project data, creating and maintaining test assets, running tests, and examining run evidence.

• 8 predefined prompts for common planning, authoring, debugging, review, refactoring, and reporting workflows.

• 9 BugBug Skills that guide the agent through safe, BugBug-specific workflows.

• Client configuration for the selected AI client.

Supported AI clients

Client	--agent value
Cursor	cursor
Claude Code	claude
VS Code	vscode
Codex	codex
GitHub Copilot	copilot

For a Streamable HTTP MCP client not listed here, follow the manual MCP configuration. The CLI plugin route remains the recommended setup because it installs both MCP and BugBug Skills.

Use it safely

Tell the agent what it may change and what evidence it should inspect. For example:

Diagnose a failed checkout run

In the BugBug project for <application>, inspect the latest failed checkout run <run-id>. This is read-only: inspect the failed-step details, screenshots, logs, and DOM evidence. If the run ID or required access is missing, ask me for it. Return the likely cause, supporting evidence, and the smallest proposed fix; do not modify the test.

For more practical prompts, see How to use AI with BugBug?.

Alternatives

Use BugBug MCP only if you only need live BugBug tools. Use BugBug Skills only if your agent already has a BugBug connection or you only need workflow guidance.

Troubleshooting

1. Re-run npx @bugbug-io/cli plugin --agent=<name> for the selected client.

2. Restart the client after installation if it does not load the plugin immediately.

3. Complete the client’s MCP OAuth sign-in flow and verify that the client can connect to https://mcp.bugbug.io/mcp.

4. Confirm that your BugBug plan includes MCP access.

Related docs

• How to use AI with BugBug?

• BugBug MCP

• BugBug MCP tools

• BugBug Skills

FAQ

What is the BugBug Plugin, and how is it different from BugBug MCP and BugBug Skills?

BugBug Plugin is the recommended way to connect an AI coding agent to BugBug. It installs BugBug MCP, BugBug Skills, predefined workflow prompts and the appropriate client configuration together. MCP gives the agent live BugBug tools, while Skills teach it how to use those tools safely and effectively.

Which AI coding agents does the BugBug Plugin support?

BugBug Plugin supports Cursor, Claude Code, VS Code, Codex and GitHub Copilot. Other AI tools that support Streamable HTTP MCP can connect to the BugBug MCP server through manual configuration, although the Plugin is the recommended setup for supported clients.

How do I install the BugBug Plugin?

Run the BugBug CLI command with the name of your AI client:

npx @bugbug-io/cli plugin --agent=<agent>

Replace <agent> with cursor, claude, vscode, codex or copilot. Restart the client if necessary, then complete the BugBug OAuth sign-in process.

How do I verify that the BugBug Plugin is connected correctly?

Ask the agent to list the names of your available BugBug projects without changing anything. If it returns the expected projects, the MCP connection and authentication are working. If it cannot access them, restart the client, repeat the OAuth flow and confirm that your plan includes MCP access.

Do I need a paid BugBug plan to use the BugBug Plugin?

BugBug Skills are available on all BugBug plans, but live access to BugBug projects through MCP requires a Pro or higher plan. Because the Plugin installs both components, users on plans without MCP access can use the Skills but cannot perform live project actions through BugBug MCP.