Adovo

Applied AI

A LangGraph batch run can skip the review interrupt

Check the LangGraph JavaScript SDK batch payload before relying on interruptBefore or other run options as a human review gate.

November 8, 2026

A LangGraph JavaScript SDK issue reports that runs.createBatch() sends camelCase run options unchanged while the server expects snake_case. In the reproduction, interruptBefore is silently ignored and the run passes a review node. The issue also lists checkpoint and durability options that may be dropped. A batch that finishes successfully can therefore violate the workflow the caller intended.

Inspect the request contract

The SDK reference defines batch creation, and its run payload reference describes interrupt options. Test the exact installed SDK against a controlled server and inspect the outgoing body. An option present in TypeScript is not proof that the wire request preserved it.

Keep approval outside an unproven path

Until the batch path passes a regression test, route high-impact actions through a path whose interrupt behavior you have verified. Put a server-side or application-level hold before irreversible steps so a missing SDK option cannot become an unreviewed action. A UI label saying “awaiting approval” is insufficient if the run has already crossed the gate.

Check the fix in a release

The issue links a fix pull request, but that is not a released-package guarantee. Pin the SDK version, test one batch item with interruptBefore, and assert that the run pauses at the intended node. Repeat for any other option your operation depends on.

Adovo’s view

Human review is a control, not a parameter name. A control exists only when the running system proves the pause happened. Test the wire contract before scaling an agent from one run to a batch.

Continue exploring

Applied AI

LangGraphAgent Reliability

A LangGraph cached task fails before it can run on mixed map keys

Diagnose a LangGraph cache-key TypeError caused by unorderable mapping keys and use a collision-safe key contract.

Read more

Applied AI

LangGraphAgent Reliability

A LangGraph checkpoint lookup can stall a shared Redis

Investigate Redis KEYS calls in LangGraph checkpoint retrieval when agent latency grows with the whole keyspace.

Read more

Applied AI

LangGraphAgent Reliability

A LangGraph Postgres checkpoint read can scan old blob versions

Investigate checkpoint read latency that grows with a thread’s history in the JavaScript Postgres saver.

Read more
Explore applied AI services →