Struct ExecutorContextMetadata 

pub struct ExecutorContextMetadata {

    pub tool_metadata: Option<HashMap<String, Value>>,
    pub metadata: Option<Value>,
    pub additional_attributes: Option<AdditionalAttributes>,
    pub external_tools: Option<Vec<ExternalTool>>,
    pub definition_overrides: Option<DefinitionOverrides>,
    pub dynamic_sections: Option<Vec<PromptSection>>,
    pub dynamic_values: Option<HashMap<String, Value>>,
    pub browser_session_id: Option<String>,
    pub env_vars: Option<HashMap<String, String>>,
    pub dry_run: Option<bool>,
    pub runtime_mode: RuntimeMode,
    pub is_sandbox: bool,
    pub tags: Option<HashMap<String, String>>,
    pub trace_context: Option<TraceContext>,
}

Metadata sent by clients (CLI, browser SDK, etc.) alongside A2A messages. Deserialized by the server to configure execution context. This is the canonical schema — all clients should serialize this struct.

Fields

tool_metadata: Option<HashMap<String, Value>>

Additional context for tools to use, passed as meta in tool calls

metadata: Option<Value>

additional_attributes: Option<AdditionalAttributes>

external_tools: Option<Vec<ExternalTool>>

External tools that delegate execution to the client

definition_overrides: Option<DefinitionOverrides>

Definition overrides supplied by the client

dynamic_sections: Option<Vec<PromptSection>>

Dynamic prompt sections to inject into the template per-call

dynamic_values: Option<HashMap<String, Value>>

Dynamic key-value pairs available in templates per-call

browser_session_id: Option<String>

Browser session ID for browser tool integration

env_vars: Option<HashMap<String, String>>

Environment variables passed from the client for execution

dry_run: Option<bool>

When true, unsafe tools are simulated via LLM instead of executed

runtime_mode: RuntimeMode

Runtime environment. Determines which system agent variants to use.

is_sandbox: bool

When true, the caller is running inside a distri sandbox container (gated on DISTRI_IN_SANDBOX=1 in the sandbox entrypoint). Server copies this onto ExecutorContext.is_sandbox so tools and prompts can detect sandbox-context execution.

tags: Option<HashMap<String, String>>

Arbitrary caller-supplied tags. Recorded on the agent span (as a JSON object under distri.tags) and merged into the thread’s attributes.

trace_context: Option<TraceContext>

Inbound distributed-trace context. When present, the agent’s root span and all descendants inherit this trace_id / parent_span_id.

Trait Implementations

impl Clone for ExecutorContextMetadata

fn clone(&self) -> ExecutorContextMetadata

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for ExecutorContextMetadata

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

Formats the value using the given formatter.

impl Default for ExecutorContextMetadata

fn default() -> ExecutorContextMetadata

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ExecutorContextMetadata

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for ExecutorContextMetadata

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.