CEROS , an AI Trust Layer built by Beyond Identity, sits directly on developers’ machines alongside Claude Code to give security teams visibility and control over an agent that operates outside traditional identity and network controls. According to The Hacker News, Claude Code can read files, execute shell commands, call external APIs, and connect to MCP servers using the full permissions of the developer who launched it, with no audit trail captured by existing security tools.
Ceros provides real-time visibility, runtime policy enforcement, and a cryptographic audit trail of every action the agent takes, including a Conversations view that shows every session and the tool calls behind it, and a Tools view with Definitions and Calls outlines. It records complete device context in under 250 milliseconds before a session starts and ties the session to a verified human identity via Beyond Identity’s hardware-bound key.
The platform also supports runtime policies, MCP server allowlisting, and signed Activity Logs to meet compliance needs, including references to SOC 2 CC8.1, FedRAMP AU-9, HIPAA, and PCI-DSS v4.0; Ceros is available now and free to get started.