My safety guard protected 2 tools and trusted the other 20 (opens in new tab)
I maintain an MCP server that lets a coding agent drive your real, logged-in Safari — the same browser where your bank, your email, and your half-written Slack messages live. The whole premise only works if there's one ironclad rule: The agent may only touch tabs it opened. Never yours. I wrote that guard early. It was a small function: before any page-mutating action, check that the target tab is one the agent owns. I dropped it into the wrapper that safari_click and safari_fill both flow th...
Read the original article