Struct MontyRuntime 

pub struct MontyRuntime { /* private fields */ }

Available on crate feature codeact-monty only.

A Python CodeRuntime for the CodeActAgent, backed by the Monty interpreter.

Build one with MontyRuntime::new for sensible defaults, or MontyRuntime::builder to set resource limits, grant OS access (mounted paths, an environment map, the host clock), or extend the language briefing. Hand the result to CodeActAgentBuilder::runtime.

Example

use std::sync::Arc;
use adk_codeact_monty::MontyRuntime;

let runtime = Arc::new(MontyRuntime::new());
// CodeActAgent::builder().runtime(runtime)...

Implementations

impl MontyRuntime

pub fn new() -> MontyRuntime

Create a runtime with conservative default resource limits suitable for untrusted, LLM-generated code: 5s wall-clock per advance, 256 MiB memory, and Monty’s recursion guard.

Relax or tighten these with MontyRuntime::builder; remove them entirely (trusted scripts only) with MontyRuntimeBuilder::unlimited.

pub fn builder() -> MontyRuntimeBuilder

Start building a runtime with custom limits or prompt additions.

Trait Implementations

impl CodeRuntime for MontyRuntime

fn start( &self, script: &str, script_name: &str, ) -> Result<RunStep, RuntimeError>

Parse and begin executing script, running until the first external call, completion, or error.

fn resume( &self, snapshot: &[u8], with: ResumeWith, ) -> Result<RunStep, RuntimeError>

Restore a suspended continuation (from PendingCall::dump) and resume it with the given value or error.

fn capabilities(&self) -> RuntimeCapabilities

Report what this runtime can do and which language/environment it accepts.

fn render_tools(&self, tools: &[Arc<dyn Tool>]) -> String

Render the tool catalog for the system prompt, in this runtime’s language and conventions.

impl Default for MontyRuntime

fn default() -> MontyRuntime

Returns the “default value” for a type.