Struct CompositeExecutor 

pub struct CompositeExecutor<A: ToolExecutor, B: ToolExecutor> { /* private fields */ }

Chains two ToolExecutor implementations with first-match-wins dispatch.

For each method, first is tried first. If it returns Ok(None) (i.e. it does not handle the input), second is tried. If first returns an Err, the error propagates immediately without consulting second.

Use this to compose a chain of specialized executors at startup instead of a dynamic Vec<Box<dyn ...>>. Nest multiple CompositeExecutors to handle more than two backends.

Tool definitions from both executors are merged, with first taking precedence when both define a tool with the same ID.

Example

use zeph_tools::{
    CompositeExecutor, ShellExecutor, WebScrapeExecutor, ShellConfig, ScrapeConfig,
};

let shell = ShellExecutor::new(&ShellConfig::default());
let scrape = WebScrapeExecutor::new(&ScrapeConfig::default());
let executor = CompositeExecutor::new(shell, scrape);
// executor handles both bash blocks and scrape/fetch tool calls.

Implementations

impl<A: ToolExecutor, B: ToolExecutor> CompositeExecutor<A, B>

pub fn new(first: A, second: B) -> Self

Create a new CompositeExecutor wrapping first and second.

Trait Implementations

impl<A: Debug + ToolExecutor, B: Debug + ToolExecutor> Debug for CompositeExecutor<A, B>

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

Formats the value using the given formatter.

impl<A: ToolExecutor, B: ToolExecutor> ToolExecutor for CompositeExecutor<A, B>

fn set_skill_env(&self, env: Option<HashMap<String, String>>)

Forward the active skill’s env injection to BOTH inner executors.

The base ToolExecutor::set_skill_env is a no-op, so without this override the composition tree built in agent_setup silently swallows env injection — the underlying ShellExecutor never sees GITHUB_TOKEN etc. Each layer ignores the call if it does not own a skill_env slot; layers that do (e.g. ShellExecutor) update their state. See #3869.

fn set_effective_trust(&self, level: SkillTrustLevel)

Forward the active skill’s trust level to BOTH inner executors.

Mirrors Self::set_skill_env: without this override, TrustGateExecutor never observes a non-Trusted level when composed under CompositeExecutor, leaving quarantine enforcement effectively bypassed. See #3869.

async fn execute(&self, response: &str) -> Result<Option<ToolOutput>, ToolError>

Parse response for fenced tool blocks and execute them.

async fn execute_confirmed( &self, response: &str, ) -> Result<Option<ToolOutput>, ToolError>

Execute bypassing confirmation checks (called after user approves).

fn tool_definitions(&self) -> Vec<ToolDef>

Return the tool definitions this executor can handle.

async fn execute_tool_call( &self, call: &ToolCall, ) -> Result<Option<ToolOutput>, ToolError>

Execute a structured tool call. Returns Ok(None) if call.tool_id is not handled.

fn is_tool_retryable(&self, tool_id: &str) -> bool

Whether the executor can safely retry this tool call on a transient error.

fn is_tool_speculatable(&self, tool_id: &str) -> bool

Whether a tool call can be safely dispatched speculatively (before the LLM finishes).

fn execute_tool_call_confirmed( &self, call: &ToolCall, ) -> impl Future<Output = Result<Option<ToolOutput>, ToolError>> + Send

Execute a structured tool call bypassing confirmation checks.

fn requires_confirmation(&self, _call: &ToolCall) -> bool

Return true when call would require user confirmation before execution.