Check if this constraint set allows the given resource type.

Compare *data* against a stored snapshot, creating it on first run. On the first run the snapshot file is written so tests pass and produce a

Audit a repo's compliance with its governance tier requirements.

Policy gate before a graph node runs. Raises: PolicyViolationError: If the node is blocked by policy.

Initialize execution context and check cancellation.

Build the 'session' sub-command parser.

Execute *func* through the circuit breaker. Sync callables return their result directly. Awaitable results are returned as a coroutin

Check if sponsor can grant a specific capability. Supports exact match, wildcard (``*``), and prefix wildcard matching. Args:

Capture a point-in-time snapshot of agent state. Args: agent_id: Agent identifier agent_state: Compl

Check abort conditions. Returns True if experiment should stop.

Check if tool call count exceeds limit.

Return ``True`` if a cascading failure is detected.

Run **all** registered checks and return a full report.

Check a requirements.txt file for supply chain risks.

Check if an agent's ring allows access to a specific resource type. Args: agent_ring: The agent's current execution ring.

Check if *value* exceeds resource limits. Returns (is_anomaly, score).

Check if a tool definition changed since registration (rug pull). Returns: An ``MCPThreat`` if a rug pull is detected, else ``Non

Check if a tool is allowed by policy.

Create a child context linked to this context.

Clear all entries (use with caution).

Create a test client for the base app.

Check files for policy violations.

Inspect a single session in detail.

Install git pre-commit hooks for Agent OS.

Sign a plugin's agent-plugin.yaml manifest.

Show the status of the Agent OS kernel.

Compare two configuration dicts and return findings for each diff. Args: source_config: Reference configuration. targ

Compare current fingerprints against a saved baseline.

Compare trust scores, flagging differences beyond *tolerance*. Args: source_scores: Reference trust scores keyed by agent id.

Run compensation (rollback) for all committed steps in reverse order. Cancellation semantics on timeout match ``execute_step``: see

Compute deterministic hash of receipt.

Compute an initial trust score for a newly registered plugin. Heuristics: - Base score: 500 (``standard`` floor). - +100 if the manifest

Configure logging from the AGENTOS_LOG_LEVEL environment variable.

Minimal mock EventEmitter matching vscode.Event<void> pattern.

Create a FastAPI app with standard health/metrics endpoints.

Create an escrow for a task. Locks credits from requester until task completion.

Convenience function to create a governed A2A agent. Args: control_plane: Agent Control Plane instance agent_id: Agent I

Create an OpenAI Agents SDK compatible guardrail callable. The returned object is an async callable with the guardrail signature ``(c

Create a Mute SQL Agent configuration This agent: - Only executes SELECT queries - Returns NULL for anything outside this capabi

Remove all data for an agent. Args: agent_did: The agent's DID identifier.

Return a denial for exceeding the configured tool-call limit.

Compare two traces and return differences. Basic comparison: checks for missing/extra spans and output mismatches.

Disconnect from the broker.

Convert GovernanceEvents to AuditEntry and write to backend.

-- Persistence ------------------------------------------------------------

Compute cost estimate for a model+task pair.

Evaluate a set of changed files against this policy. Args: files: List of DiffFile objects representing the changes. Ret

Evaluate a brainstorming conversation. Args: turn_count: Number of turns in the conversation context

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

Run all registered extractors against context in place.

Extract declared capabilities from a manifest dict. Capabilities are derived from skill names and MCP server tool lists. Args: data:

Return True if the policy at ``policy_path`` applies to the action. Args: policy_path: Path to the governance file (unused today, accepte

Filter a list of capabilities based on the plugin's trust tier. Capabilities with a higher minimum-tier requirement than the given *tier* are

Format scan results as deterministic JSON.

Format security findings for CLI output.

Convert a CloudEvent to an AMB Message. Args: event: CloudEvent to convert type_prefix: Prefix to strip from event type

Deserialize delegation from dictionary.

Extract trace context from HTTP headers.

Get an agent by ID.

* Get the dependency graph as a list of agent nodes.

Return agent IDs whose circuits are currently OPEN.

Return the number of calls made by *agent_id*. Args: agent_id: Agent identifier to inspect. Returns: The num

Return the namespace name an agent belongs to. Args: agent_did: DID of the agent to look up. Returns: The na

Retrieve alerts, optionally filtered. Thread-safe.

Returns audit log entries with optional filtering. Args: filters: Optional dict with keys ``agent``, ``action``,

Retrieve the audit log