Simulate an agent refusing a delegated task.

Delete a message. Args: message_id: Message ID Returns: True if deleted, False

Delete state.

Delete a trace.

Deny a pending tool call by its call_id. Returns True if the call was pending and is now denied.

Return a denial for a tool-argument blocked-pattern match.

Return a denial for a tool that matches a blocked tool pattern.

Return a denial for a confidence score below the configured threshold.

Valid deregistration signature for the given DID.

Remove this agent from Nexus.

Remove an agent and release all its kernel resources. Cleans up the agent's context, signal dispatcher, and VFS instance. No-op if th

Scan for orphan candidates across the fleet.

Detect the user's intent from their query. Args: query: The user's first query verbose: Print detect

Detect PCI patterns (credit card numbers).

Detect source type from document/section metadata and content. Args: document: Document to analyze s

* Discover peers advertising a given capability. Returns [] if no * registry is configured.

Discover all agent configurations in a repository. Looks for: - .agents/agents.md - .agents/AGENTS.md - agents.md (root) - A

Dispatch a parsed session command to the appropriate handler.

Enable or disable shadow mode for all executions. In shadow mode, actions are simulated but not actually executed.

End a session. Args: session_id: Session ID to end. success: Whether the session was successful. end_stat

* Ensure data directory exists.

Initiate a trust handshake with a peer agent. Creates a cryptographic challenge, records the handshake, and returns a signed token the peer c

Estimate savings from Anthropic caching. Returns: Tuple of (tokens_saved, cost_saved_usd)

Evaluate whether an agent's changes are within scope. Decision logic: 1. ``config.mode == "off"`` → always ``PASS``.

Evaluate current metrics and make rollout decision. Returns: Decision dictionary with action, reason, and next p

Evaluate and auto-fail the task if trust check fails. Returns the TrustResult. If denied, the envelope is transitioned to FA

Public access to the AGT ``input`` intervention point evaluation.

Public access to the AGT ``input`` intervention point evaluation.

Record a warning without running policy checks.

Simple exact match evaluator.

Execute branches sequentially (Public Preview).

Get the execution order of agents based on delegation links.

Export metrics in Prometheus text format. Returns: Prometheus-formatted metrics text

Write all current metrics into a Prometheus-compatible exporter. Args: exporter: Any object implementing :class:`PrometheusExport

Extract a TraceContext from incoming headers. Use on the *server* side to correlate incoming spans with the caller's trace.

Fetch satellite data for location (simulated)

Recursively find files by name, skipping junk directories.

Finish the current span.

Flush the batch queue.

* Format a Date as a locale time string (HH:MM:SS). * * @param date - The date to format. * @returns Formatted time string.

Format findings as a human-readable report. Args: findings: List of SecurityFinding objects. Returns: Markdown-formatted rep

Format alert as PagerDuty Events API v2 payload.

Return a human-readable table of profiling results.

Return a CLI-friendly progress bar, e.g. ``[████████░░] 82% (8,200/10,000 tokens)``.

Format scan results as plain text suitable for terminals and tests.

Create a PolicyBundle from a configuration dictionary. Args: data: Raw configuration dict (typically loaded from YAML).

Create from AZURE_TENANT_ID + AZURE_CLIENT_ID env vars.

Import an AgentIdentity from a JWK Set. Args: jwks: A dict representing a JWK Set. kid: Optional key ID to filter by. If None, us

Translate the generic :class:`SandboxConfig` into a :class:`HyperlightConfig`. Resource limits (``memory_mb``, ``timeout_seconds``) b

Parse a CausalTraceId from its string representation.

Convert a TrustedAgentCard to an AI Card. Bridges the existing AgentMesh ``TrustedAgentCard`` format to the cross-protocol AI Card st

Deserialize policy from YAML string (minimal parser, no PyYAML dep).

Load a PolicyDocument from a YAML file.

Generate a new verification identity.

Generate a new random X25519 key pair.

* Generate agent code from spec

* Generate GitHub Actions workflow YAML

Return the stored session for *session_id*, if present.

* Get all available policies.

* Get all containment events.

* Get all available frameworks