Struct SwlRuntime 

pub struct SwlRuntime { /* private fields */ }

SWL Workflow Runtime

Implementations

impl SwlRuntime

pub fn new(client: Arc<ApiClient>) -> Self

Create a new runtime with API client

pub fn with_tool_registry( client: Arc<ApiClient>, tool_registry: Arc<ToolRegistry>, ) -> Self

Create a new runtime with a custom tool registry

pub fn new_dry_run() -> Self

Create a dry-run runtime (no actual LLM calls)

pub async fn with_file_persistence( client: Arc<ApiClient>, workflow_name: &str, ) -> Result<Self>

Create a runtime with file-based state persistence

pub fn with_max_tool_iterations(self, max: usize) -> Self

Set the maximum number of tool iterations per agent execution

pub async fn execute_workflow( &self, doc: &SwlDocument, workflow_name: &str, inputs: HashMap<String, VarValue>, ) -> Result<ExecutionResult>

Execute a workflow from a document with state persistence

pub async fn persist_state(&self) -> Result<()>

Persist the current execution context state

pub async fn load_state(&self) -> Result<()>

Load state from persistence

pub async fn get_context(&self) -> ExecutionContext

Get execution context

pub async fn get_execution_trace(&self) -> Vec<ExecutionEvent>

Get the execution trace (telemetry events)

pub async fn get_telemetry_summary(&self) -> WorkflowTelemetry

Get aggregated telemetry summary

pub async fn export_telemetry_json(&self) -> Result<String>

Export telemetry data as JSON string

pub async fn clear_telemetry(&self)

Clear all telemetry data

pub async fn get_agent_trace(&self, agent_name: &str) -> Vec<ExecutionEvent>

Get the raw execution trace for a specific agent

Trait Implementations

impl Clone for SwlRuntime

fn clone(&self) -> Self

Faithful clone: shares the API client, tool registry, and execution context (so trace events and state written by the clone are visible to the original), and preserves dry_run / max_tool_iterations / workflow_name. The safety checker is rebuilt from the same config.

GuardedSwlRuntime relies on this when spawning parallel agent tasks — cloning used to substitute a fresh dry-run runtime, so guarded parallel and map-reduce workflows returned [DRY-RUN] placeholder strings while reporting Completed.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.