Struct ShellSkill 

pub struct ShellSkill { /* private fields */ }

Shell execution skill with production-grade command validation.

Commands are parsed into segments (split on shell metacharacters) and each segment’s base command is validated against the configured CommandPolicy. Additionally, a set of unconditionally dangerous patterns is always blocked regardless of policy configuration.

Implementations

impl ShellSkill

pub fn new() -> Self

Create a new ShellSkill with the default blocklist policy.

pub fn with_policy(policy: CommandPolicy) -> Self

Create a new ShellSkill with a custom CommandPolicy.

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

Set the maximum number of bytes to capture from stdout.

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

Set the maximum number of bytes to capture from stderr.

Trait Implementations

impl Default for ShellSkill

fn default() -> Self

Returns the “default value” for a type.

impl Skill for ShellSkill

fn descriptor(&self) -> &SkillDescriptor

Return the skill’s metadata descriptor.

fn validate_arguments( &self, call: &ToolCall, permissions: &PermissionSet, ) -> ArgentorResult<()>

Validate that the specific arguments in this tool call are permitted by the given permission set. Override for skills that need argument-level checks. Default: always returns Ok(()) (no argument-level validation).

fn execute<'life0, 'async_trait>( &'life0 self, call: ToolCall, ) -> Pin<Box<dyn Future<Output = ArgentorResult<ToolResult>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Execute the skill with the given tool call.