wterm brings WebMCP to browser terminals

This is a meaningful step toward structured computer use: terminal automation becomes an explicit tool contract instead of brittle DOM scraping or simulated keyboard events.

• –Per-terminal opt-in gives developers a practical permission boundary for multi-terminal pages
• –Read-only mode can enable safer inspection and debugging workflows before granting input access
• –Listing, reading, typing, and key presses cover the core primitives agents need to coordinate shell sessions
• –Framework support lowers adoption friction across modern frontend stacks
• –WebMCP remains experimental, so browser support and API stability are the main near-term caveats