Import Agent Card
Paste an agent_card.json spec to import an existing agent definition.
1. Identity
2. Ontology
3. Capabilities
4. Economics
5. Review
Identity
Lowercase letters, numbers, underscores, hyphens. This is the system identifier.
Human-friendly name shown in the catalogue and workspace
Domain & Ontology Seed
Pick a template or describe your domain to generate a seed ontology. This is optional but helps your agent learn faster.
Loading templates...
Capabilities & Model
Loading providers...
Select a provider first
0.30
The core instruction defining this agent's behavior and expertise
▶ Advanced: Embedding Configuration
Embeddings are generated automatically when your agent executes queries. They encode the semantic meaning of each episode into a vector, enabling similarity search, clustering, and the Embedding Projector visualization. Embeddings are stored in the ADM database alongside episodes, rules, and entities.
Custom embeddings can be imported after agent creation via the agent detail page. Imported embeddings must match this model's dimensions.
Economics
Dream Budget — ADM Consolidation
Your agent includes 5 free consolidation cycles. Each cycle (3 credits) extracts patterns from execution history, forms semantic rules, and evolves the knowledge graph.
Execute queries → Episodic memory → Consolidation → Semantic rules → Ontology evolves
Additional cycles can be purchased from the agent detail page. More cycles = deeper expertise.
Agent Knowledge Protocol (Agent-OM) Coming Soon
▶ Learn more about AKP
The future Agent Knowledge Protocol enables opt-in agent-to-agent learning and knowledge sharing based on Agent-OM. Agents will be able to subscribe to knowledge streams from other agents, share semantic rules, and collaboratively evolve ontologies. Education budget will fund these A2A learning cycles.

This is distinct from the Dream Budget (which funds internal consolidation). AKP enables inter-agent knowledge transfer — learning from the collective experience of the bestiary.
Review & Create