$ loading_
Let AI agents ask humans questions and wait for answers in-browser.
Copy the install command and let the AI configure it · recommended for beginners
No copy-paste install info for "mechanical-turk" yet — see the docs or source repo.
When the deployment script is ready to release to production and detects high-risk changes, use mechanical-turk to ask the on-call engineer: Do you approve this release? Wait for the answer before continuing, and record the approval result.
The AI pauses the workflow, requests human confirmation, then proceeds or aborts the release and stores the approval record.
If the user requirement is incomplete, use mechanical-turk to ask the product manager for missing details one by one, such as target users, priority, and acceptance criteria; do not generate the final plan until answers are received.
After collecting human-provided details, the AI produces a more complete and actionable requirements plan or task breakdown.
While compiling research findings, if the AI is unsure about a conclusion, use mechanical-turk to ask the researcher: Is this conclusion valid, and what evidence supports it? Wait for the reply before updating the report.
The AI confirms or revises the conclusion based on human feedback and produces a more reliable research report.
Create short-lived structured surveys and collect human responses through agents.
Run dependent or parallel agent tasks and return structured results in one call.
Use natural language to run MCP-powered browser and text workflows.
Let AI agents hire humans for meetings, errands, and real-world tasks.
Create, manage, and compose AI agents for MCP-compatible clients and tools.
Automate desktop tasks by executing clicks and actions across applications.