Show HN: Mantis – AI agents browse 36% faster than with Playwright MCP (github.com)

A new AI browsing agent called Mantis has been released, showcasing a remarkable performance boost over Playwright's MCP. Mantis allows AI agents to navigate the web using concise text input rather than cumbersome screenshots, thereby enhancing browsing speed by approximately 36%. With a simple two-command installation, Mantis integrates seamlessly with models such as Claude Code, Codex, and Cursor, making it accessible for various user setups including Windows and Linux servers. This development is significant for the AI/ML community as it reduces the number of input tokens and API calls needed for web interaction, leading to enhanced efficiency. Benchmark results indicate that Mantis processes tasks in about half the time while consuming 3.5 to 5 times fewer tokens compared to traditional methods. This efficiency could greatly optimize AI's ability to interact with web content, which is critical for applications in data analysis, information extraction, and automated web tasks. However, users should be aware of security concerns, particularly regarding untrusted page content and login protections.