Struct DefaultSkillRuntime 

pub struct DefaultSkillRuntime { /* private fields */ }

Default skill runtime — LLM prompt injection + tool-calling loop.

Implementations

impl DefaultSkillRuntime

pub fn new(registry: Option<Arc<dyn SkillRegistry>>) -> Self

pub fn with_permissions( self, allowed_network: bool, allowed_paths: Vec<PathBuf>, ) -> Self

Trait Implementations

impl Debug for DefaultSkillRuntime

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl SkillRuntime for DefaultSkillRuntime

fn execute<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, skill_id: &'life1 str, input: &'life2 str, context: &'life3 ExecutionContext, ) -> Pin<Box<dyn Future<Output = Result<SkillOutput, SkillError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Execute a skill: injects the skill prompt into the LLM context, runs the tool-calling loop, and returns the final output.

fn execute_tool<'life0, 'life1, 'async_trait>( &'life0 self, tool_name: &'life1 str, args: Value, ) -> Pin<Box<dyn Future<Output = Result<Value, SkillError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Execute a single named tool directly (bypasses LLM prompt injection).

fn validate_permissions<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, skill: &'life1 Skill, _context: &'life2 ExecutionContext, ) -> Pin<Box<dyn Future<Output = Result<(), SkillError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Validate that the execution context satisfies all permissions required by the skill.

fn preflight_check<'life0, 'life1, 'async_trait>( &'life0 self, skill: &'life1 Skill, ) -> Pin<Box<dyn Future<Output = Result<Vec<PreflightWarning>, SkillError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Run preflight checks on a skill before it is enabled or executed. Returns a list of warnings (non-blocking) or errors (blocking).