flowlens

FlowLens MCP server lets your coding agent (Claude Code, Cursor, Copilot, Codex) access a recorded browser session to debug and test with full context. The server works with the FlowLens browser extension to capture user actions, network activity, console output, storage, and DOM events, and then exposes that recording to your agent through the MCP interface. Once configured in your MCP client, your agent can inspect the flow and reason about issues without having to reproduce them from scratch, speeding up debugging and regression testing.

To use FlowLens with MCP, install the server (via pipx) and add the MCP configuration to your client. The server exposes a standard stdio transport so your agent can communicate with it as part of its normal MCP workflow. If you’re sharing flows or connecting to the FlowLens platform, you can provide a FLOWLENS_MCP_TOKEN in your config to enable access to shareable flows and tokens.

Typical usage scenarios include bug reporting with full browser context, regression testing by recording critical user flows and generating test scripts, and sharing captured flows with teammates for collaborative debugging. The tools available via this MCP server focus on translating a recorded session into actionable insights for your coding agent without requiring manual reproduction of issues every time.

Prerequisites:

• Python-installed environment (recommended Python 3.8+)
• pipx installed (for isolated Python tools)
• FlowLens Chrome extension installed and configured if you plan to record flows locally

1. Install the FlowLens MCP server via pipx:

pipx install flowlens-mcp-server

2. (Optional) Upgrade to the latest version:

pipx upgrade flowlens-mcp-server

3. Run or verify the server (it exposes an MCP endpoint on stdio when used via the MCP client):

flowlens-mcp-server

4. Add the MCP configuration to your client (example shown in repository README):

"flowlens": {
  "command": "flowlens-mcp-server",
  "type": "stdio",
  "env": {
    "FLOWLENS_MCP_TOKEN": "YOUR_FLOWLENS_MCP_TOKEN"
  }
}

Prerequisite notes:

• Ensure your FlowLens token is securely stored and not committed to public repos.
• If you prefer running via pipx run syntax, you could alternatively use: pipx run flowlens-mcp-server in place of the direct command, depending on your environment.

Tips and caveats:

• The FLOWLENS_MCP_TOKEN enables access to shareable flows via the FlowLens platform; obtain it from the FlowLens setup wizard if you need to connect to shareable flows.
• The MCP config uses stdio transport; ensure your MCP client supports stdio and that the FlowLens server command is resolvable in your environment.
• If you encounter connectivity issues, verify that the FlowLens extension is installed and that the token (if provided) is correct. Check environment variable scoping in your shell or container if running inside a sandbox.
• You can omit the env block if you don’t need FlowLens token-based features, but token-based sharing unlocks additional capabilities for collaboration.