Display an overview of the red team dataset

Display expected results

Show the long-term benefits of the upgrade purge strategy.

Display the benchmark methodology

Show how to run the full benchmark

Display sample prompts from each category

Flush in-flight events and release resources.

Stop the server.

Flush and shut down exporters.

Create a sibling span (same parent, different operation).

Send SIGCONT — resume kernel operations after SIGSTOP.

Sign data with this identity's private key.

Return base64-encoded Ed25519 signature.

Sign *data* using the acquired key. Raises: KeyAcquisitionError: If the handle has expired.

Sign a DMZ data handling policy to receive access.

Build and sign a reasoning attestation envelope.

Sign an SBOM and return an envelope with the signature. The envelope wraps the SPDX payload together with a :class:`SignatureBundle`

signedHeaderResolver verifies a signed agent-identity header set of the form: X-Agent-Id: <agent_id> X-Agent-Timestamp: <unix-seconds> X-Agent-Signatu

Simulate transaction execution In production, this would: 1. Fork chain state using Ganache/Anvil 2. Execute

Simulate the WebSocket emitting an error event mid-stream.

Simulate a brainstorming conversation that's too short. This should be evaluated as FAILURE (too short = not creative enough).

Simulate a brainstorming conversation with deep exploration. This should be evaluated as SUCCESS (many turns = deep discussion).

Simulate a LangChain RAG agent processing a task.

Simulate a LangChain RAG agent processing a single task. Drives the AgentSRECallback through the same lifecycle events that a real LangChain

Simulate production log stream.

Simulate the complete registration flow.

Simulate a pharmaceutical cold chain shipment with mixed readings.

TODO: Mutate one audit entry to simulate log tampering, then call audit.verify_integrity() and print whether the tamper was detected.

Simulate a troubleshooting conversation. This should be evaluated as SUCCESS if resolved quickly (<= 3 turns).

Simulate a troubleshooting conversation where user gets trapped. This should be evaluated as FAILURE (> 3 turns = trapped, not engaged).

Simulate agent becoming unhealthy (for testing)

Total number of registered identities.

Get current size of DLQ.

Return list of skill IDs.

Get all SLIs provided by this integration.

Sort entries by timestamp descending and cap at MAX_DISPLAY.

* Split SQL on top-level semicolons (semicolons outside of quoted strings). * Trailing empty/whitespace segments are dropped.

* Start the REST server and wait for it to become healthy. * * @returns Result with endpoint URL if successful

Start the agent

Start the MCP server.

Start the worker.

Start the worker.

Start a span representing delegation between agents. Args: tracer: OpenTelemetry tracer instance. from_did: DID of the delegating

Start the Prometheus HTTP metrics server. Args: port: TCP port to listen on. Defaults to ``9090``.

Start a span representing an LLM inference call. Args: tracer: OpenTelemetry tracer instance. model_name: Name of the model (e.g.

Start the monitoring agent in Ghost Mode.

Start a span representing a policy check. Args: tracer: OpenTelemetry tracer instance. policy_name: Name of the policy being eval

Start a span representing a tool call. Args: tracer: OpenTelemetry tracer instance. tool_name: Name of the tool being invoked.

Demo: The Startup Opportunity Show how this becomes a product/SDK.

Convenience function for one-shot stateless execution. Creates an ephemeral ``StatelessKernel`` and ``ExecutionContext``, executes the action

* Perform basic static analysis to find issues

* Perform basic static analysis to find issues

Step 1: Create the Control Plane

Step 2: Create an Agent

Step 3: Execute a Simple Action

Step 4: Check the Audit Log

Step 5: Try an Action Without Permission

Break a trace into interactive steps (root spans with their children).

Stop the agent gracefully