Question: Does BrowserMCP solve the issue with huge MCP responses? #139
Description
I'm using Claude Code + Playwright MCP and often get huge tool responses (full page dumps). This can cause the MCP response text to be trimmed, so Claude never sees the full page. It's also slow and wastes context.
Idea: Have MCP first return a short page summary (main sections, key elements, node IDs), and then let the LLM call another tool like get_node_details(node_id) to fetch detailed content only for that part of the page.
Does BrowserMCP support anything like this, or could it? Or will it make the whole process even slower?