loopuman-human-tasks
от seesayearn-boop
Productivity & TasksКратко
Give your AI agent access to real human workers.
Описание
Что делает этот навык
Loopuman routes agent tasks to human workers for judgment, verification, or real-world actions that automated tools can't handle reliably. Workers are paid instantly in cryptocurrency. Agents can block and wait for a response or post tasks asynchronously and poll later.
It lets agents delegate tasks requiring human judgment without building or managing a separate workforce pipeline.
Когда его использовать
- Checking if a product review is genuine or AI-generated
- Moderating user-submitted content before it goes live
- Fact-checking claims in agent-generated reports
- Labeling images or text for machine learning datasets
- Transcribing audio files that automated tools get wrong
Пример сценария
User asks: Is this product review genuine or fake? 'Amazing product, changed my life, 5 stars!'
- 1
Receives the review text to evaluate
- 2
Calls ask_human with a clear task description and sets budget_cents to 50
- 3
Waits for a human worker to read and respond (up to 5 minutes by default)
- 4
Returns the worker's reasoning and verdict to the user
Human worker flags the review as likely fake, citing its generic language, lack of product specifics, and common fake review patterns
Требования
- LOOPUMAN_API_KEY environment variable (free registration via API call)
- Funded Loopuman account balance (minimum $0.10 per task, via USDC/USDT/cUSD on Celo network or Stripe)
Установка
clawhub install seesayearn-boop/loopuman-human-tasksВыполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.