Anthropic Turns Claude Security into Claude Code Plugin
Anthropic has developed the deep scanning capability of Claude Security into a Claude Code plugin, which is now available in beta. Paid users of Claude Code can install it, while the hosted version is still only available to Enterprise customers.
Users can scan entire codebases, branch differences, PRs, or individual commits, and after discovering vulnerabilities, they can generate patches for self-review and application. The plugin runs multiple agents within a session for independent verification, with usage counted towards consumption.
This extension is event-driven for development tools, directing funds towards AI security scanning and code tools, benefiting Claude Code users while traditional standalone security scanning tools face pressure.
Source: Public Information
ABAB AI Insight
Claude Security entered public testing for Enterprises on April 30, 2026, previously available for limited preview under the name Claude Code Security, now being made accessible to a broader range of paid users through a plugin format, while maintaining a high-end positioning for the hosted version.
The capital pathway reflects the embedding of security capabilities into daily coding workflows, motivated by the aim to lower the threshold for vulnerability discovery and enhance developer retention, while also reducing false positives through a multi-agent architecture.
Compared to GitHub Copilot's security features and other AI code tools, Anthropic is in the stage of transitioning from enterprise hosting to widespread terminal plugin adoption.
Essentially, this is a technological substitution, where the mechanism involves multi-agent collaboration within sessions making deep scanning a natural part of the coding process, rather than a separate post-factum step.
ABAB News · Cognitive Laws
Pluginization is the shortest path for capability embedding
Multi-agent verification reduces false positive costs
Security scanning is becoming a part of daily coding