Loading...
MCP-Human provides a human-in-the-loop capability for AI assistants by creating Mechanical Turk tasks to answer questions, enabling access to human knowledge.
Boost this tool
Subscribe to listing upgrades or segmented pushes.
MCP-Human provides a human-in-the-loop capability for AI assistants by creating Mechanical Turk tasks to answer questions, enabling access to human knowledge.
MCP-Human is relatively safe in sandbox mode for testing. However, using it in production requires careful management of AWS credentials and monitoring of MTurk costs. The reliance on an external form introduces a dependency and potential security concern.
Performance is limited by the time it takes for human workers to complete tasks on MTurk. Response times can vary depending on the reward amount and the complexity of the task.
The primary cost is the reward amount paid to human workers on MTurk. There may also be minor costs associated with using the MTurk API.
{
"mcpServers": {
"human": {
"command": "npx",
"args": ["-y", "mcp-human@latest"]
}
}
}AWS_ACCESS_KEY_IDAWS_SECRET_ACCESS_KEYaskHumanAsks a question to a human worker on Mechanical Turk and retrieves the answer.
Creates tasks on MTurk, incurring costs and potentially exposing sensitive questions.
checkHITStatusChecks the status of a previously created HIT and retrieves any submitted assignments.
Read-only operation that retrieves information about existing HITs.
mturk-accountProvides access to MTurk account information like balance and HITs.
Provides read-only access to MTurk account information.
Environment Variable
MCP-Human is relatively safe in sandbox mode for testing. However, using it in production requires careful management of AWS credentials and monitoring of MTurk costs. The reliance on an external form introduces a dependency and potential security concern.
Autonomy is limited by the need for human input and the polling mechanism for retrieving results. The level of autonomy can be configured by adjusting the reward amount and HIT validity period.
Production Tip
Start with a low reward amount and gradually increase it until you get satisfactory response times. Monitor MTurk costs closely to avoid unexpected expenses.
AmazonMechanicalTurkFullAccess is recommended, or a custom policy with specific MTurk permissions.
Set the `DEFAULT_REWARD` environment variable or specify the `reward` parameter in the `askHuman` tool.
Set the `MTURK_SANDBOX` environment variable to `true` for sandbox mode and `false` for production mode.
The HIT will expire after the specified validity period. You can check the status of the HIT using the `checkHITStatus` tool.
No, custom HTML/JS is not supported. The form is a static HTML page hosted on GitHub Pages.
Avoid asking questions that involve sensitive or confidential information. MTurk is not designed for handling sensitive data.
Increase the reward amount or make the question more clear and concise.