Struct MontyPythonCodeTool 

pub struct MontyPythonCodeTool { /* private fields */ }

Available on crate feature tools only.

Embedded Python code execution tool backed by the Monty interpreter.

The LLM-facing description is composed, not authored: a short base contract followed by the executor’s own prompt_snippet, which describes mode semantics, granted filesystem roots, environment variable names (never values), clock availability, the no-network/no-subprocess guarantee, and registered host functions. Because both the description and the in-interpreter behavior derive from the same configuration, they cannot drift.

In REPL mode, interpreter sessions are keyed by the full ADK session identity — app name, user id, and session id — so state never leaks between users even when session id strings repeat across users. The session map is bounded by an LRU cap (default 100); an evicted session’s next call transparently starts a fresh interpreter.

For container-backed CPython (pip packages, C extensions, the complete standard library), use PythonCodeTool instead.

Requires the code-embedded-python feature. Without it, the tool returns a structured "rejected" result.

Required Scopes

Returns ["code:execute"]. No elevated scopes needed.

Example

use adk_tool::{MontyPythonCodeTool, Tool};

let tool = MontyPythonCodeTool::new();
assert_eq!(tool.name(), "monty_python_code");
assert_eq!(tool.required_scopes(), &["code:execute"]);

Implementations

impl MontyPythonCodeTool

pub fn new() -> MontyPythonCodeTool

Available on crate feature code only.

Create a one-shot tool with a fully sandboxed interpreter: no filesystem, no environment variables, no clock, no host functions.

pub fn repl() -> MontyPythonCodeTool

Available on crate feature code only.

Create a REPL-mode tool with a fully sandboxed interpreter. Variables, functions, and imports persist across calls within an ADK session.

pub fn builder() -> MontyPythonCodeToolBuilder

Available on crate features code-embedded-python and code only.

Start configuring a tool with OS-access grants, host functions, and limits (forwarded to MontyExecutorBuilder).

Example

use adk_code::PathAccess;
use adk_tool::MontyPythonCodeTool;

let tool = MontyPythonCodeTool::builder()
    .allow_path("/data", "/srv/agent/data", PathAccess::ReadOnly)
    .environ_var("PROJECT", "acme")
    .system_clock()
    .build_repl()?;

Trait Implementations

impl Default for MontyPythonCodeTool

fn default() -> MontyPythonCodeTool

Returns the “default value” for a type.

impl Tool for MontyPythonCodeTool

fn parameters_schema(&self) -> Option<Value>

The schema is built from configuration: reset exists only in REPL mode, where it discards the persistent session before executing.

fn response_schema(&self) -> Option<Value>

The response envelope is fixed across both modes (and the feature-off fallback), so the schema is declared for the model. Script-level failures are data in this envelope, never a ToolError.

fn name(&self) -> &str

Returns the unique name of this tool.

fn description(&self) -> &str

Returns a human-readable description of what this tool does.

fn required_scopes(&self) -> &[&str]

Returns the scopes required to execute this tool.

fn is_read_only(&self) -> bool

Indicates whether this tool performs no side effects.

fn is_concurrency_safe(&self) -> bool

Indicates whether this tool is safe for concurrent execution.

fn execute<'life0, 'async_trait>( &'life0 self, ctx: Arc<dyn ToolContext>, args: Value, ) -> Pin<Box<dyn Future<Output = Result<Value, AdkError>> + Send + 'async_trait>>

where 'life0: 'async_trait, MontyPythonCodeTool: 'async_trait,

Executes the tool with the given context and arguments.

fn declaration(&self) -> Value

Returns the tool declaration that should be exposed to model providers.

fn enhanced_description(&self) -> String

Returns an enhanced description that may include additional notes. For long-running tools, this includes a warning not to call the tool again if it has already returned a pending status. Default implementation returns the base description.

fn is_long_running(&self) -> bool

Indicates whether the tool is a long-running operation. Long-running tools typically return a task ID immediately and complete the operation asynchronously.

fn is_builtin(&self) -> bool

Indicates whether this tool is a built-in server-side tool (e.g., google_search, url_context).