Experimental

Use Experimental only when a native agent integration does not cover the workflow. Native hooks are more reliable because MCP focus signals depend on the agent remembering to call a tool.

CLI task tracking

Enable CLI Task Runner & Status Tracking to make these commands available:

• b2v run — track any shell command.
• b2v task — create a script-owned session.
• b2v status — report progress to that session.

Leave this off unless you use those commands.

Local MCP server

1. Confirm Server Binary shows Ready.
2. Copy the generated configuration for Claude or Cursor.
3. Add it to the configuration file shown in the app.
4. Restart the agent and verify that it can see the back2vibing MCP tools.

Do not add MCP configuration to agents that already show Active under Agents.

Remote access

Remote access lets a web-based agent reach the local MCP server through a tunnel.

1. Enable Remote Access.
2. Copy the generated remote MCP configuration into the web agent.
3. Test one focus signal.
4. Disable remote access when it is no longer needed.

The tunnel URL is a secret that points to your local machine. Do not publish or share it.

Experimental features may change or fail. Keep a native or local workflow available as a fallback.