Dock

Dock enables multiple users and AI agents to interact with shared data in real-time through a cloud-based interface. It employs WebSocket connections for live updates, ensuring that all changes are instantly reflected across all clients. This architecture allows for seamless collaboration, as both human users and AI agents can read and write to the same state simultaneously without conflicts.

Dock implements OAuth 2.1 with dynamic client registration, allowing clients to self-register and obtain access tokens without manual intervention. This approach streamlines the authentication process, enabling users to authorize once while Dock manages token refresh and revocation seamlessly. The architecture supports both user-based and API key-based access, enhancing flexibility for various use cases.

Dock supports structured data manipulation through a table mode that includes various typed columns such as text, number, status, and more. This allows users to create, update, and delete rows of data with specific types, ensuring data integrity and type safety. The implementation uses a schema-based approach to enforce data types and relationships, making it easier for users to manage complex datasets.

Dock maintains an audit trail for each AI agent, capturing all actions taken by the agent along with timestamps and user identifiers. This feature is implemented using a logging mechanism that records changes to the workspace, allowing users to track the history of interactions and modifications made by agents. This transparency is crucial for accountability and debugging.

Dock tracks recent events in workspaces, allowing users to retrieve a history of actions such as row creations, updates, and deletions. This capability is implemented through an event sourcing pattern, where each change is logged as an event that can be queried later. Users can access this event history to understand the evolution of their workspace and to facilitate collaboration.