Return the default ω for this level.

Define a capability for an agent. Implements CapabilityValidatorInterface.define_capability()

Define a capability for an agent. Args: agent_id: ID of the agent capability_name: Name of the capab

Define a capability for an agent. Args: agent_id: ID of the agent capability_name: Name of the capab

Delegate to a child agent with narrowed capabilities. The child agent's capabilities MUST be a subset of the parent's. This

Delete key.

Delete key.

Delete key.

Delete key.

Delete a file.

Delete an episode by ID. Returns True if found and deleted.

Delete a message.

Delete a message file.

Delete *session_id* and return ``True`` when it existed.

Remove a credential. Returns True if it existed.

Deregister an agent.

* Delete agent.

Remove a checkpoint image.

Delete a document. Args: document_id: The document ID Returns: Deletion status

Delete an agent's keypair from the HSM. Args: agent_id: Agent identifier. Raises: KeyError: If no keypair ex

Delete a policy

Delete a file from the Virtual File System. Removes a file from the shared project state.

Demo 1: Basic Lifecycle Management Shows the new API for registering and managing agents.

Demo 2: Circuit Breaker Pattern Shows how circuit breakers prevent cascading failures.

Demo: Firewall enforces cost limits.

Demo: The key insight - AI can be creative (high temperature) because the firewall provides safety.

Demo: Firewall blocks dangerous file operations.

Demo: Firewall blocks dangerous SQL operations.

Demo 5: Dependency Graph Shows enforced startup order based on dependencies.

Demo: Firewall restricts email domains.

Demo 9: Full Integration Shows all features working together.

Demo 6: Graceful Shutdown Shows preservation of in-flight operations during shutdown.

Demo 3: Health Monitoring Shows liveness and readiness probes.

Demo 7: Agent Observability Shows metrics collection and Prometheus export.

Demo 8: Resource Quotas Shows resource limiting per agent.

Demo: Firewall approves safe operations.

Demo 4: Horizontal Scaling Shows dynamic scaling of agent replicas.

Deny a pending escalation request (called by human/UI).

An evaluator that always denies.

Mark a tool version as deprecated. Args: name: Tool name. version: Version to deprecate. message: Optional deprecation me

Remove an agent from the registry.

Deregister an agent. Requires Ed25519-Timestamp auth and the caller's DID must match the DID being deregistered — only the ho

Human-readable description.

Deserialize from bytes.

Tear down the sandbox and release resources.

Detect potential bias in text content. This is a simplified implementation. In production, would use ML models train

Scan a single text for prompt injection.

Scan multiple texts for prompt injection.

Detect PII in text.

Detect agents approaching rate limits

Detect agents with repeated failures