Under Section 3 "Use of our Services," the Terms specifically state:
Except when you are accessing our Services via an Anthropic API Key or where we otherwise explicitly permit it, to access the Services through automated or non-human means, whether through a bot, script, or otherwise.
This is listed as one of the prohibited ways to access or use the Services.
This means that:
You are not allowed to use automated means (bots, scripts, etc.) to access the web interface (Claude.ai)
If you want to access Claude through automation, you need to use the official Anthropic API with a proper API key
You would need to follow the Commercial Terms of Service when using the API
But this is just using the Model Context Protocol that Anthropic itself released to let Claude use tools in order to do actions outside the context of the conversation.
If this code were used to automate interactions with Claude.ai (such as automatically sending prompts, clicking buttons, or extracting responses), it would be accessing the service "through automated or non-human means" which is explicitly prohibited by the Terms of Service.
https://www.anthropic.com/legal/consumer-terms
Under Section 3 "Use of our Services," the Terms specifically state:
Except when you are accessing our Services via an Anthropic API Key or where we otherwise explicitly permit it, to access the Services through automated or non-human means, whether through a bot, script, or otherwise.
This is listed as one of the prohibited ways to access or use the Services. This means that:
You are not allowed to use automated means (bots, scripts, etc.) to access the web interface (Claude.ai) If you want to access Claude through automation, you need to use the official Anthropic API with a proper API key You would need to follow the Commercial Terms of Service when using the API
But this is just using the Model Context Protocol that Anthropic itself released to let Claude use tools in order to do actions outside the context of the conversation.
The code has func like:
initialize_browser click_element input_text scroll_down send_keys
If this code were used to automate interactions with Claude.ai (such as automatically sending prompts, clicking buttons, or extracting responses), it would be accessing the service "through automated or non-human means" which is explicitly prohibited by the Terms of Service.
There's a puppeteer MCP server in the official GitHub repo by Anthropic which gives Claude access to similar tools too: https://github.com/modelcontextprotocol/servers/tree/main/sr...
it is weird that https://github.com/jspahrsummers is a contributor there. Maybe he started this without asking Anthropic legal team?