CVE-2026-45134

Affected versions of the LangSmith SDK package are vulnerable to Deserialization of Untrusted Data due to improper handling of prompt manifests fetched from the LangSmith Hub. The `pull_prompt` and `pull_prompt_commit` methods in Python, as well as `pullPrompt` and `pullPromptCommit` in JS/TS, deserialize prompt manifests that may contain serialized LangChain objects and model configurations controlled by external parties. An attacker can exploit this by publishing a malicious prompt to the LangSmith Hub, which, when pulled by `owner/name`, can lead to Server-side Request Forgery (SSRF), prompt injection, or behaviour manipulation by altering application configurations with attacker-supplied arguments.