mcp-human
If you are the rightful owner of mcp-human and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.
MCP-Human is a Model Context Protocol server that enables AI assistants to get human input via Amazon Mechanical Turk.
MCP-Human is a Model Context Protocol (MCP) server designed to facilitate human-in-the-loop AI systems by enabling AI assistants to request human input when needed. It achieves this by creating tasks on Amazon Mechanical Turk, allowing real humans to answer questions posed by AI systems. This server is primarily a proof-of-concept, demonstrating the potential of integrating human feedback into AI workflows using the MCP standard. The server operates in sandbox mode by default for testing purposes, with the option to switch to production mode for real requests. It requires AWS credentials with MTurk permissions and can be configured using various environment variables. The architecture consists of an MCP server that integrates with MTurk and a static HTML form for human workers to complete tasks. The server supports simple text-based questions and answers, with limitations such as single assignment per HIT and simple polling for results.
Features
- Human-in-the-loop AI: Integrates human feedback into AI workflows using Amazon Mechanical Turk.
- Sandbox and Production Modes: Supports testing in sandbox mode and real requests in production mode.
- Configurable Environment: Allows configuration through environment variables for flexibility.
- AWS Integration: Requires AWS credentials with MTurk permissions for operation.
- Simple Text-Based Interaction: Supports simple text-based questions and answers.
Tools
askHuman
Allows an AI to ask a question to a human worker on Mechanical Turk.
checkHITStatus
Check the status of a previously created HIT and retrieve any submitted assignments.
