What problem does it solve? AI agents lack structured, persistent memory: facts about people, projects, tasks, and events end up scattered across conversations with no way to query, relate, or validate them. This Skill provides a typed knowledge graph stored as an append-only JSONL file so agents can remember, link, and verify information across sessions and skills. ## Core Features & Use Cases - Typed Entity Management: Create, query, update, and delete entities such as Person, Project, Task, Event, Document, and Credential via a Python CLI. - Relations & Graph Traversal: Link entities with typed relations (has_owner, blocks, part_of) and query dependencies in outgoing, incoming, or both directions. - Schema Constraint Validation: Enforce required properties, enums, forbidden fields, cardinality, acyclic dependency checks, and Event end >= start rules defined in a YAML schema. - Cross-Skill Shared State: Other skills declare ontology read/write contracts so commitments, tasks, and actions flow between skills through one graph. - Use Case: Tell the agent "remember Alice owns the Website Redesign project" and later ask "what tasks depend on task X?" — the graph answers with validated, queryable structure. ## Quick Start Ask the agent to remember a new person named Alice and create a project owned by her, then list all people stored in the ontology.
