Call A Human

Acerca de

Call A Human provides AI agents with a critical 'pause button,' allowing them to solicit human input or explicit approval before proceeding with potentially destructive or irreversible operations. By integrating with various communication channels like Slack, Telegram, or macOS system dialogs, it ensures that an AI agent cannot, for instance, delete a production database without first obtaining a human's go-ahead. The system works by blocking the AI's execution until a human responds to its query or approval request, providing a crucial layer of human oversight in AI-driven workflows and enhancing the safety and reliability of AI-powered tasks.

Características Principales

• Receive human approval for irreversible AI actions via request_approval tool.
• Supports multiple notification channels: Slack, Telegram, and native macOS system dialogs.
• Compatible with MCP-enabled AI clients such as Claude Desktop, Cursor, and Windsurf.
• Provides an append-only JSONL audit log for all human interaction events.
• 1 GitHub stars
• Gather human-provided information and context using the ask_human tool.

Casos de Uso

• Gathering necessary human context or preferences when an AI agent faces an ambiguous decision or needs clarification.
• Integrating a human-in-the-loop mechanism for sensitive or high-impact automated AI workflows.
• Preventing AI agents from executing destructive commands or making critical changes without explicit human consent.