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
fix(memory): a finished plan's notes are queued under the caller's profile scope
sanil-23ed8b3197d196
diff --git a/crates/openhuman-core/src/memory/lifecycle/plan_notes.rs b/crates/openhuman-core/src/memory/lifecycle/plan_notes.rsindex 5ff7c56fba..d03a6564a7 100644--- a/crates/openhuman-core/src/memory/lifecycle/plan_notes.rs+++ b/crates/openhuman-core/src/memory/lifecycle/plan_notes.rs@@ -76,10 +76,13 @@ pub async fn learn_plan_notes();let notes = learnings(plan);let total = notes.len();- // The queue does blocking file I/O, as it does for the memory tool.- let runtime = tokio::runtime::Handle::current();- let queued = tokio::task::spawn_blocking(move || {- runtime.block_on(async {+ // The queue opens its storage under the current profile scope. Like the+ // memory tool's writes, the work runs on a task that carries the caller's+ // scope (the session host calls this inside the profile's scope): a bare+ // spawn would run unscoped, and in SaaS the queue refuses every write.+ let dispatch = crate::core::runtime::DispatchCtx::ambient();+ let queued = dispatch+ .spawn(async move {let mut queued = 0;for (text, kind) in notes {let args = json!({@@ -105,9 +108,8 @@ pub async fn learn_plan_notes(}queued})- })- .await- .unwrap_or(0);+ .await+ .unwrap_or(0);tracing::debug!(thread_id = %plan.thread_id,queued,