Disclosed weakness · no CVE recorded

Claude Code could skip a pre-tool check or lose its approval request

Anthropic · Claude Code

First documented disclosure · Review checkpoint

Summary

Anthropic fixed two cases where a hook intended to check a tool action could lose its request for user approval or be skipped.

Severity and exploitation evidence

No CVSS score recorded in this review.

Exploitation not established: The release documents the correction. It does not identify a victim or malicious exploitation.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

The client could lose an approval requirement while combining hook decisions or handling an MCP-backed hook response.

Conditions required

A deployment relies on concurrent pre-tool checks or an MCP-backed hook to review a tool action.

Potential impact

An expected security check could be absent from the decision. The release does not establish that any particular downstream action succeeded.

Affected versions and configurations

Anthropic did not identify the full affected version range.

Fix and mitigation status

Anthropic lists the fix in Claude Code 2.1.321.

Practical next steps

Update to a release containing the 2.1.321 fixes. Anthropic says the corrected client refuses a tool call when it cannot safely process the hook response.

Operating lesson

Keep approval requirements in force when a secondary control fails or its answer cannot be processed.

Evidence and limits

The vendor lists the two cases separately in the same release. This grouped entry concerns enforcement of pre-tool checks and does not assume a shared root cause.

Sources