Struct Metadata 

pub struct Metadata {

    pub session_id: Option<String>,
    pub agent_id: Option<String>,
    pub parent_agent_id: Option<String>,
    pub is_subagent: bool,
    pub is_delegated_work: bool,
    pub agent_kind: Option<String>,
    pub agent_role: Option<String>,
    pub task_id: Option<String>,
    pub task_kind: Option<String>,
    pub turn_id: Option<String>,
    pub session_final: Option<bool>,
    pub correlation_id: Option<String>,
    pub extra_metadata: Option<BTreeMap<String, String>>,
    pub http_headers: Option<BTreeMap<String, String>>,
    pub wire_format: Option<WireFormat>,
}

Correlation and routing metadata attached to a request or response.

All fields are optional (or default-empty); algorithms and observers use whichever are present (e.g. to key per-session state or emit correlated telemetry). The agent-lineage fields (parent_agent_id, is_subagent, agent_kind, agent_role, task_kind, turn_id, session_final) are populated for requests from a coding agent. extra_metadata is a free-form escape hatch for host-specific keys.

Fields

session_id: Option<String>

Stable id for a multi-request session/conversation.

agent_id: Option<String>

Id of the agent making the request.

parent_agent_id: Option<String>

Id of the parent agent, when this request comes from a child agent.

is_subagent: bool

Whether the harness identified this request as coming from a child agent.

is_delegated_work: bool

Whether this request carries delegated sub-agent work and should be routed to the sub-agent target. Computed from raw harness signals only, independent of Self::agent_kind, which may be set by an unrelated operator label (x-switchyard-agent-kind).

agent_kind: Option<String>

Harness-defined kind of agent call, such as collab_spawn or review.

agent_role: Option<String>

Semantic agent role, such as explorer, worker, or reviewer.

task_id: Option<String>

Id of the task the request belongs to.

task_kind: Option<String>

Semantic task class supplied by the harness.

turn_id: Option<String>

Id of the current agent turn.

session_final: Option<bool>

Whether the harness signalled this is the session’s final request (e.g. the host may evict per-session state). None when the harness said nothing.

correlation_id: Option<String>

External trace/request id for joining with the host’s telemetry.

extra_metadata: Option<BTreeMap<String, String>>

Arbitrary host-defined key/value metadata.

http_headers: Option<BTreeMap<String, String>>

HTTP headers to attach when forwarding the request/response, if any.

wire_format: Option<WireFormat>

The wire format the request/response was originally encoded in, if known.

Implementations

impl Metadata

pub fn from_headers(headers: &BTreeMap<String, String>) -> Self

Normalizes harness-specific request headers into correlation metadata.

Explicit x-switchyard-* headers win. Compatible host headers can populate correlation and session metadata, but do not by themselves mark a request as delegated work. Structured turn metadata is preferred over compatibility projections. A request with a non-empty x-claude-code-agent-id is treated as a child agent (its parent inferred from the session when not stated). Sub-agent routing status is taken from an explicit x-switchyard-is-subagent header when present, and otherwise inferred from native agent lineage or structured harness signals.

pub fn is_subagent_work(&self) -> bool

Whether this request should be routed to the sub-agent target.

Returns self.is_delegated_work, which is computed in parse_sub_agent from raw harness signals only — independent of agent_kind, which may be populated by an unrelated operator label (x-switchyard-agent-kind).

Trait Implementations

impl Clone for Metadata

fn clone(&self) -> Metadata

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Default for Metadata

fn default() -> Metadata

Returns the “default value” for a type.