pub struct GetRunResponse {Show 20 fields
pub run_id: String,
pub tenant_id: String,
pub agent_id: String,
pub session_id: Option<String>,
pub status: RunStatus,
pub input: Option<Map<String, Value>>,
pub output: Option<Map<String, Value>>,
pub metrics: Option<RunMetrics>,
pub error: Option<String>,
pub created_at: String,
pub started_at: Option<String>,
pub completed_at: Option<String>,
pub team_run_id: Option<String>,
pub metadata: Option<Map<String, Value>>,
pub step_seq: Option<i64>,
pub artifacts: Option<Vec<Artifact>>,
pub resource_limits: Option<GetRunResponseResourceLimits>,
pub changed_files: Option<Vec<String>>,
pub pending_approvals: Option<Vec<Map<String, Value>>>,
pub pending_input: Option<GetRunResponsePendingInput>,
}Expand description
GetRunResponse model.
Fields§
§run_id: String§tenant_id: String§agent_id: String§session_id: Option<String>§status: RunStatus§input: Option<Map<String, Value>>§output: Option<Map<String, Value>>Run output. When a run is truncated by its step-budget cutoff (output.truncated === true) AND the platform has UARP_CONTINUATION_TOKEN_KEY configured, output.continuation_token carries an opaque HMAC-signed token that resumes the run via POST /runs/{id}/continue. With no key configured no token is minted and the field is absent; the token is an opaque string to every client.
metrics: Option<RunMetrics>§error: Option<String>§created_at: String§started_at: Option<String>§completed_at: Option<String>§team_run_id: Option<String>Team run ID if part of a team execution
metadata: Option<Map<String, Value>>User-supplied metadata
step_seq: Option<i64>Current step sequence number
artifacts: Option<Vec<Artifact>>Run artifacts
resource_limits: Option<GetRunResponseResourceLimits>Resource limits for the run
changed_files: Option<Vec<String>>Workspace paths this run WROTE, sorted, present only when the request carries
?changed_files=true. Recorded per (run, path) at write time, so a file rewritten three
times appears once and two concurrent tool calls cannot lose one another’s entry.
It is a record of writes, not a diff: a path the run DELETED or moved is not here, and neither is a change made by something else while the run was going. Recording is best-effort after the write has already succeeded — a failure to record is logged and leaves the list short rather than failing the edit — so treat it as “at least these” rather than proof that nothing else changed. Capped at 500 paths. Rows expire 30 days after the run.
pending_approvals: Option<Vec<Map<String, Value>>>Tool calls the run is blocked on, taken from the most recent run.awaiting_approval event.
ABSENT — not empty — when the run is not awaiting approval, and absent too if the scan
fails, which is deliberate: a failed scan must not turn a readable run into an error.
pending_input: Option<GetRunResponsePendingInput>The question the run is blocked on, taken from the most recent run.awaiting_input event
(runs.ts:673-681). Like pending_approvals it is ABSENT rather than empty when the run is
not awaiting input. A squad chat reads this to render the prompt; the document never
mentioned it, so a client written from the document alone showed a blocked run as merely
running.
Trait Implementations§
Source§impl Clone for GetRunResponse
impl Clone for GetRunResponse
Source§fn clone(&self) -> GetRunResponse
fn clone(&self) -> GetRunResponse
1.0.0 (const: unstable) · Source§fn clone_from(&mut self, source: &Self)
fn clone_from(&mut self, source: &Self)
source. Read more