Get all events for a specific conversation. Args: conversation_id: The conversation identifier

Return a human-readable digest of batched alerts.

Query audit entries with optional filters. Args: start_time: Filter entries after this time end_time: Filter

Get revocation details for an agent, or None if not revoked.

Get an audit entry by its unique ID.

Get execution history

Get the health status of an agent

Get event history, optionally filtered. Args: limit: Maximum number of events to return (most recent first)

Returns agents ranked by trust score (descending). Args: limit: Maximum number of leaderboard entries.

Get an agent's manifest by DID.

Get recorded metrics

Get a specific pattern by ID.

Get imported records, optionally filtered by eval name.

Return the full AVP reputation profile for an agent. Includes score, confidence, tier, risk factors, attestation count, and EigenTrus

Get statistics about the review queue.

Get or create the capability scope for an agent. Args: agent_did: The agent's decentralized identifier. Returns:

Get current risk score for an agent. Creates a default score if one does not already exist. Args: agent_did: The agent's

Return all SLI values as a dict (for recording into Agent-SRE SLO).

Convert imported evaluations to SLI-compatible values. Returns dict mapping SLI type names to lists of float values.

Get the startup order using topological sort. Returns agents in order such that dependencies are started first. Rais

Get the complete file system state. Returns: VFSState object representing the entire file system

Get statistics about the conversation. Returns: Dictionary with statistics

Get validator statistics.

Get cache statistics. Returns: Dictionary with cache statistics

Return a snapshot of concurrency statistics. Thread-safety: YES — reads are guarded by the thread lock.

Get runner statistics.

Return deduplication statistics.

Return MCP resource templates.

Acquire an access token from the GCE metadata server.

Return cumulative token usage statistics. Returns: A dict with keys ``prompt_tokens``, ``completion_tokens``, ``total

Return current trust score (0–1000) for an agent.

Get current usage for an agent

Return current budget status for *agent_id*.

Return all collected violations.

Return all collected violations.

Translate a v4 :class:`GovernancePolicy` to an AGT manifest dict. Args: policy: A v4 ``GovernancePolicy`` (or any object with the

Execute a two-layer governance check before allowing an agent action. Layer 1 — AGT Policy: Is this agent authorized for this action? La

Grant this session read access to another session's data. Only effective under READ_COMMITTED isolation. Args: other_ses

Return an async callable guarded by input/output intervention points.

Handle an agent failure and attempt recovery. Returns the new agent instance if recovery succeeds, None otherwise.

Handle MCP prompts/list request.

Handle MCP resources/read request.

Check if principal has a specific role.

Check if agent has a specific Entra API scope.

Check if storage backend is healthy.

Get hash field value.

Import a single evaluation result. Accepts Phoenix/Arize evaluation format: - eval_name: Name of the evaluation (e.g. "hallu

Start a stdio MCP server, initialize it, and enumerate advertised primitives.

Invoke a prompt with governance. This is for direct chat/completion calls. The prompt is evaluated at the AGT ``input`` inte

Check if this trace is an ancestor of another (same trace tree).

Check if attestation is present and not expired.

Return True if neither clock causally precedes the other.

Check if this identity has expired. Returns: True if expired, False if still valid or no TTL set

Check if the challenge has exceeded its time-to-live.

Check if escrow has expired.

Check if this entry has expired.

Check if this alert should fire.

Check if an action has any reversibility.

Quick check: is this action safely reversible?

Return whether the wrapper is in a stopped (SIGSTOP) state.

Check if credential is valid (active and not expired). Returns: True if status is active and the credential has not expired.

Governed async kickoff.

Return sorted list of registered adapter names.

List plugins visible to an organization (org-specific + global). A plugin is visible to an organization when its ``organization`` fie

Return all credential handle names (admin operation; not for agents).

Return manifests for all installed plugins. Returns: List of installed plugin manifests.

List all registered org policies.

Convenience: register a Cedar backend. Args: policy_path: Path to a ``.cedar`` policy file. policy_content: Inline Ce

Load metadata manifest from file.

Load all YAML trust policies from a directory. Scans for ``*.yaml`` and ``*.yml`` files in sorted order. Args: directory: Path to th

Add a log entry to the span.

Create tamper-evident audit entry.

Log quality feedback for a Helicone-tracked request. Args: helicone_id: Helicone request ID rating: Positive (True) o

Log a shadow mode execution (simulated, not real). Shadow mode allows the governance layer to return a plausible simulated result to

CLI entry point.

Create a SOC2 governance policy scoped to a specific role.

Get the compliance mapping for an action type. Args: action_type: Action identifier (e.g. ``"data_access"``). Returns: