Anthropic Claude Update

Enterprise plans can enable skill and plugin security scanning in Claude to automatically check third-party uploads and edits for malicious content. The feature is in beta and returns pass, warn, or fail results before a skill or plugin can run.

• Adds security scanning for third-party skills and plugins to detect malicious content before they run in an organization.
• Blocks skills and plugins flagged as malicious; blocked items cannot currently be overridden by users or admins.
• Scanning runs in an isolated environment and keeps only the result and basic metadata after completion.

• Beta skill and plugin security scanning for Enterprise plans.
• Pass, warn, and fail scan outcomes with a caution banner for warn results.
• Organization-wide enablement with optional control through custom role capabilities.

• Skill and plugin scanning does not apply to existing skills and plugins already in the organization before the feature is enabled.
• Skills created with Claude, connected MCP server skills, MCP servers, and hooks are not scanned.
• Organizations using CMEK, ZDR, or HIPAA configurations do not get the added scan.
• Admins cannot currently approve a blocked skill for the organization.

• Enable the feature in Organization settings > Skills as an Owner or Primary Owner.
• Turn on the Skill and plugin security scanning capability for custom roles that should use scanning.
• Uploading the same skill again without changes returns the same scan result; fixes must address the flagged content.
• Scanning is cached, so re-uploading the same skill can return a result almost instantly.

Product Information

Vendor: Anthropic

Product: Claude

Product type: Other

Platform: Web

Vendor information published: Aug 6, 2026

Original release notes: View on vendor site

Published on updatealert.io: Aug 8, 2026

Description: AI assistant for drafting, analysis, coding, and enterprise workflows.