ontology

You can share state across different agent skills by using a typed knowledge graph to structure agent memory. This enables entity linking and allows multiple skills to access and update shared ontology objects during execution.

The best way to represent tasks and projects is by using a typed knowledge graph for entity management. You can create, query, and update entities like Tasks and Projects while defining their properties and enforcing relationship constraints.

To enforce constraints on entity relationships in structured agent memory, you can define required properties within the typed knowledge graph schema. The graph then validates these constraints during entity CRUD operations and multi-step graph transformations.

Yes, you need Python to manage JSONL file operations for knowledge graph storage. The system relies on Python scripts to handle graph storage, execute schema modifications, and perform entity CRUD operations via the command line.

Yes, you can plan multi-step actions as graph transformations within composable skills. The knowledge graph supports planning these transformations while enforcing constraints and maintaining shared state across the execution sequence.

You need a typed knowledge graph for structured agent memory when your application requires entity linking, constraint enforcement, and state sharing across multiple skills, which simple state variables cannot adequately structure or manage.