The vulnerability arises from unsafe deserialization of untrusted data in langsmith-sdk's prompt pull methods (pull_prompt / pull_prompt_commit in Python and pullPrompt / pullPromptCommit in JS/TS). Prior to versions 0.8.0 (Python) and 0.6.0 (JS/TS), the SDK fetched prompt manifests from the LangSmith Hub without distinguishing whether the prompt was from the user's own organization or a public prompt controlled externally. These manifests can include serialized LangChain objects and model configurations that affect runtime behavior. This lack of validation enables potential exploitation via deserialization of malicious data. The issue is classified under CWE-502 (Deserialization of Untrusted Data) and carries a CVSS 3.1 score of 7.1, indicating high severity. The vulnerability is addressed by updating to the fixed SDK versions.

Successful exploitation could allow an attacker to influence the runtime behavior of applications using vulnerable versions of the langsmith-sdk by supplying malicious serialized objects in public prompt manifests. This may lead to partial confidentiality impact (high impact on confidentiality), limited integrity impact (low), and no availability impact. There are no known exploits in the wild as of the published date. The vulnerability affects users who pull public prompts using the affected SDK versions.

Upgrade langsmith-sdk to Python version 0.8.0 or later, or JS/TS version 0.6.0 or later, where this vulnerability is fixed. Since no official patch advisory is provided beyond these version updates, applying these version upgrades is the recommended remediation. Patch status is not explicitly confirmed beyond these version increments; check the vendor's official channels for any additional guidance.