Qeravio
Canonical AI event

The LangChain library now retries failed tool calls in createAgent.

The "createAgent" function now retries failed tool calls. Error messages from tools now include a status of "error".

27 Sept 20261 verified claims1 sources1 observations
What happened

The official source reports this update: langchain@1.5.14. Patch Changes #11712 34edee1 Thanks @blove ! - createAgent no longer ends the run when a returnDirect tool fails. A tool error, including arguments that fail the tool's schema, now goes back to the model so it can correct the call and retry; the run still ends on the first successful returnDirect result. Error ToolMessage s produced by the tool node now carry status: "error" , and a ToolMessage returned from a custom handleToolErrors function defaults to that status when it sets none.

Why it matters

This official update documents a development concerning langchain@1.5.14. Its practical significance depends on the scope and evidence stated by the source.

What to watch next

Read the official source update and verify its stated scope, evidence and timing before acting on it.

Connected knowledge

Entities affected by this event

Langchainjssubject
Continue this topic
agents-softwareOpenRouter 0.4.15 updates OpenAI dependency to 1.6.027 Sept 2026agents-softwareThe Anthropic library now accepts SystemMessages at any position in the message list.22 Sept 2026agents-softwareGoogle's LangChain adds direct routing for Bedrock and Vertex AI models.1 Sept 2026agents-softwareLangChain v1.5.13 fixes OpenAI API replay for responses with multiple reasoning items.1 Sept 2026agents-softwareLangChain AWS Adds Claude Model Variants to Supported Tool Choices1 Sept 2026
Evidence trail

Sources behind the event

LangChain.js official releaseslangchain@1.5.14
Editorial presentation

Read the public story

Open story