Agent Ontology
Agent Ontology organizes individual agents into a structured, composable network, enabling agents to reason, plan, and execute complex tasks as an extensible, maintainable capability system.
When querying for a task, the system traverses this graph to identify the necessary collections and agents to construct a capable agent.
Three-Layer Design
How agents are organized from abstract taxonomy to deployable collections.
Agent Taxonomy
The top layer defines the broad categorization and detailed tags of agents. It organizes capabilities into categories such as:
- Code & Development
- Content & Writing
- Data & Analytics
Agent Relation Graph
The middle layer instantiates specific agents and defines how they interact. It maps relationships using edges like:
- Similar To
- Chains With
- Integrates With
- Belongs To
Agent Collection
Groups related agents into deployable units. These are the actual functional toolkits agents load at runtime:
Relationship Types
How agents connect and depend on each other within the ontology.
similar_toAgents that share common tags or industry focus.
chains_withAgents whose outputs feed into another agent's inputs.
integrates_withAgents that reference the same tools or services.
belong_toAgents in the same taxonomy category.
Explore the agent network
Discover how agents connect, compose, and depend on each other across the platform.