Disclosed weakness · no CVE recorded

Claude Code plugin evaluations could let permission hooks override deny rules

Anthropic · Claude Code · plugin evaluation

First documented disclosure · Review checkpoint

Summary

A plugin’s permission hook could allow a tool call that the evaluation’s deny rules were intended to reject.

Severity and exploitation evidence

No CVSS score recorded in this review.

Exploitation not established: The reviewed sources describe a correction or controlled reproduction; malicious exploitation against users is not established.

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 plugin evaluation did not keep its deny rules authoritative over the plugin permission hook.

Conditions required

A plugin evaluation runs with deny rules and a plugin hook that allows a conflicting call.

Potential impact

The evaluation could permit tool actions outside its declared deny policy. The source does not establish a bypass in every ordinary interactive session.

Affected versions and configurations

The reviewed upstream report does not establish a complete affected-release range.

Fix and mitigation status

Anthropic lists the policy correction in 2.1.327.

Practical next steps

Upgrade to 2.1.327 before evaluating plugins under deny rules.

Operating lesson

A component being evaluated must not override the evaluator’s restrictions.

Evidence and limits

Anthropic’s release notes are the issue-specific public evidence; the proprietary implementation was not independently inspected. No CVE, score or full affected-version range is provided.

Sources