Spooky AI

Through a lightweight open-source Python library combined with a LangChain integration, developers can equip their AI agents with the ability to send real-time notifications and receive structured responses from users on their mobile devices, turning one-way automation into a genuine human-in-the-loop system.

The core workflow is elegantly simple. Developers install the Spooky Python library, create an account to receive an API key, and provide their AI agent with a LangChain tool that calls the Spooky library with a user ID and a query message.

When the agent reaches a decision point where it needs human input whether to confirm a sensitive action, choose between options, or gather additional context it triggers a push notification to the user's device.

The user responds directly from the notification, and the agent receives the answer and continues execution accordingly.

Rich notification support gives the Spooky experience a polished, professional quality beyond simple text alerts. Agents can send notifications that include images, enabling visual confirmation workflows where users approve actions based on screenshots or generated previews.

Custom response types allow developers to configure whether the agent needs a free-text response, a yes/no confirmation, a selection from predefined options, or a numerical input giving full flexibility over how different decision points are handled in the conversation between agent and human.

Spooky integrates naturally with the LangChain ecosystem, which is the most widely adopted framework for building production AI agent systems. This means teams already using LangChain for agent orchestration can add human-in-the-loop communication without switching frameworks or rebuilding existing pipelines.