Show HN: Preloop – An MCP proxy for human-in-the-loop tool approvals (preloop.ai)

Preloop has launched as a governance layer for Multi-Cloud Platforms (MCP), allowing organizations to implement approval gates for AI tool calls. This innovative MCP proxy intercepts sensitive operations such as deployments, database writes, and payments, collecting human approvals to ensure safer AI agent actions. By transforming potentially dangerous automatic actions into processes that require explicit human consent, Preloop enhances risk management while maintaining operational efficiency, as only the high-risk operations are paused for approval. This tool is particularly significant for the AI/ML community as it bridges the gap between automation and oversight, fostering accountability in AI deployments. By integrating with popular clients like Claude Code, Cursor, and Windsurf without requiring code changes, Preloop enables teams to streamline workflows and enforce compliance while reducing the chances of costly errors. Its features include support for complex approval conditions, team quorum requirements, and notifications via platforms like Slack. As the need for robust governance frameworks in AI continues to grow, Preloop offers a practical solution to address safety concerns without sacrificing the agility of development processes.