APPVIEWX has expanded its Agent Identity Security platform to improve visibility and control over AI-driven agents. The update adds Shadow AI detection, with a broader AI Bill of Materials that now includes the skills agents can access, not just models and credentials.
This enables security teams to see both sanctioned and unsanctioned agents—including browser-based and script-based shadow agents—and to assess what those skills enable, flag unsafe capabilities, and maintain a central view across endpoints, SaaS, and cloud platforms.
The release also introduces an MCP Gateway to govern all MCP servers, including shadow servers, with just-in-time access to minimise standing privileges and redaction of sensitive data. A new Agent Kill Switch allows runtime termination of any agent and its active sessions, triggered automatically by events or via policies, or on demand from the console. The capability targets both sanctioned and shadow agents and operates independently of an MCP gateway.
Additionally, AppViewX now provides expanded cost and compliance controls for AI programmes, including token usage telemetry and policy thresholds, and compatibility with regulatory frameworks such as OWASP Top 10 for Agentic Applications, MITRE ATLAS, GDPR, HIPAA, ISO, NIST SP 800-53 Rev. 5, NIST AI RMF, the EU AI Act, SOC 2, and related disclosures.