Struct MemoryToolExecutor 

pub struct MemoryToolExecutor { /* private fields */ }

Executes memory_search and memory_save tool calls on behalf of the agent.

Implementations

impl MemoryToolExecutor

pub fn new(memory: Arc<SemanticMemory>, conversation_id: ConversationId) -> Self

Create with default validator and persistent (non-ephemeral) semantics.

pub fn with_validator( memory: Arc<SemanticMemory>, conversation_id: ConversationId, validator: MemoryWriteValidator, ) -> Self

Create with a custom validator (used when security config is loaded).

pub fn ephemeral(self) -> Self

Mark this executor as ephemeral (bare mode).

When set, memory_save reports that the content is session-only and will not be available after the session ends.

pub fn with_consent_gate( self, trust_slot: MemoryConsentTrustSlot, confirm_threshold: ContentTrustLevel, ) -> Self

Attach the write-time memory-consent gate (issue #6490, MemGhost).

trust_slot must be the same MemoryConsentTrustSlot the owning Agent ratchets up in sanitize_tool_output — see AgentBuilder::with_memory_consent_trust_slot. confirm_threshold is the minimum trust tier (inclusive) that requires Channel::confirm before memory_save persists (memory.consent_gate.confirm_threshold in config).

pub fn with_audit(self, logger: Arc<AuditLogger>) -> Self

Attach an audit sink so every memory_save write is recorded with source attribution (issue #6490, MemGhost part D).

pub fn with_audit_all(self, audit_all: bool) -> Self

Set whether memory_save writes are audited, mirroring memory.consent_gate.audit_all (issue #6559). Pass the config value here regardless of consent_gate.enabled — audit attribution and the interactive/disclosure gate are independent switches, matching Agent::persist_message_inner’s background-write-path gating (store.rs).

Trait Implementations

impl ToolExecutor for MemoryToolExecutor

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

Execute bypassing the consent-gate confirmation check (called after the user approves).

memory_search has no confirmation policy of its own, so it delegates to ToolExecutor::execute_tool_call unchanged.

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

Return the tool definitions this executor can handle.

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

Parse response for fenced tool blocks and execute them.

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 requires_confirmation(&self, call: &ToolCall) -> bool

Return true when call would require user confirmation before execution.

fn checkpoint_undo(&self, _n: usize) -> CheckpointActionResult

Undo the last n checkpointed write commands.

fn checkpoint_redo(&self) -> CheckpointActionResult

Redo the last undone checkpoint.

fn checkpoint_list(&self) -> CheckpointListResult

List the current undo stack entries and redo depth.

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

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

fn execute_confirmed( &self, response: &str, ) -> impl Future<Output = Result<Option<ToolOutput>, ToolError>> + Send

Execute bypassing confirmation checks (called after user approves).

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

Inject environment variables for the currently active skill. No-op by default.

fn set_effective_trust(&self, _level: SkillTrustLevel)

Set the effective trust level for the currently active skill. No-op by default.

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

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