Apply time-based decay to trust scores.

Apply *decay_factor* to *base_score*.

AGT D1.1: apply a single-target transform payload. `transform` is the canonical mutation payload returned by AGT `transform` decisions. The r

Mark escalation as approved.

Approve an escalation request. Returns True if found and updated.

Approve a review item. Args: review_id: ID of the review to approve reviewer_notes: Optional notes f

Async query with trust verification.

Compute stroke-dashoffset for the filled portion of the arc.

Async version of run_step.

Serialise stats to a plain dictionary.

* Assess risk level based on action type.

Governed async streaming events execution.

* Attach policies to agent.

Attach a SpanSink to receive completed spans in real time.

Run a full credential audit.

Run a full credential audit.

Return an immutable snapshot of audit events.

Govern AutoGen agents in-place.

Award contracts to lowest bidders

Map remaining-percentage to a CSS color variable.

Render multiple responses at once. Args: responses: List of polymorphic responses Returns:

Pre-delegation governance check.

ADK before_tool_callback hook. Returns None to allow, or a dict with error to block.

Mark the start of a conversation turn.

Begin a new turn within the current session.

Benchmark evaluating a 100-rule policy.

Benchmark evaluating a 10-rule policy.

Benchmark an async function.

Benchmark creating and appending AuditEntry objects.

Benchmark querying audit log entries by action.

Benchmark AuditEntry field access overhead (serialization proxy).

Benchmark burn rate alert detection time.

Benchmark canary step analysis criteria evaluation.

Benchmark chaos schedule blackout-window evaluation.

Benchmark circuit breaker state check overhead.

Benchmark concurrent kernel operations.

Benchmark ErrorBudget remaining/burn-rate calculations.

Benchmark the overhead of execution time tracking.

Benchmark fault injection event creation and recording.

Benchmark full governance overhead for each adapter type.

Benchmark kernel.execute() when the action is allowed.

Benchmark kernel.execute() when the action is denied by policy.

Benchmark LangGraphKernel before_node_execution hot-path overhead. Simulates the two checks that fire on every node entry: blocked-pattern sc

Benchmark checking if a tool is in the allowed list.

Benchmark GovernancePolicy initialization (adapter startup cost).

Benchmark blocked pattern matching against tool arguments.

Benchmark rollback condition checking.

Benchmark SharedPolicyEvaluator cross-project evaluation.

Benchmark evaluating a single-rule policy.

Benchmark SLI recording throughput across all 7 indicator types.

Benchmark SLO.evaluate() throughput and latency.

Benchmark instantiation throughput across all 9 templates.

Benchmark traffic split weight calculation during rollout progression.

Benchmark loading a policy from YAML.

Block a signal. Returns False if signal is unmaskable.

Block an agent by DID.

Broadcast price signal to all agents

Log a budget warning at WARNING level.

Build/retrieve an identity for the given agent name.

Build availability snapshot from current state.

Build compliance snapshot from current state.

Build edge <line> elements, append to group, return array.

Help side panel (slides in from right).