The fastest, cheapest, most efficient open-source agent harness. Run more than 500 agents on a $10 VPS.
Rust agent-orchestrationai-agentsai-assistantdesktopllmlocal-firstmcppersonal-aiprivacyrustsecond-braintauri
feat(triage): add origin for external inbound agent turns
Steven Enamakeld3876e042247
diff --git a/crates/openhuman-core/src/triage/origin.rs b/crates/openhuman-core/src/triage/origin.rsindex 8367225fd3..2690dd0459 100644--- a/crates/openhuman-core/src/triage/origin.rs+++ b/crates/openhuman-core/src/triage/origin.rs@@ -15,6 +15,8 @@//! - [`local_trigger_origin`] — the caller and the payload are both local.//! - [`remote_trigger_origin`] — the payload arrived from outside and is//! attacker-influenceable.+//! - [`external_inbound_origin`] — a pinned agent's turn over events pushed+//! into the inbound queues; remote content, the same posture.//!//! Deliberately *not* offered: a single blanket label. A webhook body and a//! desktop notification are not the same trust proposition, and one label for@@ -69,6 +71,24 @@ pub fn remote_trigger_origin(envelope: &TriggerEnvelope) -> AgentTurnOrigin {}}+/// The origin for a pinned agent's turn over a batch of external events+/// pushed into the inbound queues (`inbound.offer`).+///+/// The same posture as [`remote_trigger_origin`]: the turn is legitimate+/// automation, but the content steering it was pushed from outside, so every+/// external-effect call parks for approval instead of running. `job_id` is+/// `inbound:<source>:<turn_id>` — the configured source name and the queues'+/// turn id, both system-generated, never event text.+#[must_use]+pub fn external_inbound_origin(source: &str, turn_id: &str) -> AgentTurnOrigin {+ AgentTurnOrigin::TrustedAutomation {+ job_id: format!("inbound:{source}:{turn_id}"),+ source: TrustedAutomationSource::Workflow {+ require_approval: true,+ },+ }+}+#[cfg(test)]#[path = "origin_tests.rs"]mod tests;