Trust3 AI has integrated its Agent DOS platform with Anthropic’s Claude Code, giving enterprise security and AI governance teams full visibility into Claude Code activity
Trust3 AI today announced an integration between Agent DOS, its agent governance platform, and Anthropic’s Claude Code. The integration gives enterprise security and AI governance teams full visibility into Claude Code activity without requiring any software installation on developer machines.
The integration covers both Claude Enterprise and Claude Team organizations. It extends the Agent DOS agent, identity, and trace model, already in use for Copilot, Foundry, M365, and Databricks, to autonomous coding agents.
Claude Code runs directly on developer machines. It reads and modifies source code while calling tools and Model Context Protocol (MCP) servers under a developer’s credentials. Until now, security and governance teams had no reliable way to determine who was using it, what systems it connected to, or what actions it performed. The alternative has been endpoint agents, which many organizations resist on security and friction grounds.
“Many enterprises are using Claude Code to write code that ships to production,” said Balaji Ganesan, CEO and Co-founder, Trust3 AI. “Agent DOS gives security complete oversight, so agent-written code ships under the same scrutiny.”
Also Read: OpenAI models are coming to Google Cloud
Agent DOS solves the problem through two complementary mechanisms. For Claude Enterprise customers, an API pull mechanism uses Anthropic’s Compliance and Analytics APIs to retrieve session transcripts, tool calls, MCP server activity, and per-user usage data.
For Claude Team customers, an admin-pushed OpenTelemetry configuration directs Claude Code to send telemetry to the customer’s own OpenTelemetry Collector and S3 bucket, which the Trust3 AI Collector then reads. Both pathways are agentless and read-only. Both feed the same Collector, Governance, and Agent DOS pipeline behind Trust3 AI’s existing platform coverage.
The integration gives technical AI and security leaders four things. A complete inventory of every Claude Code user as an accountable AI agent with a named owner. Visibility into MCP server connections and tool calls, drawn from whichever telemetry source the organization’s plan provides. Explicit surfacing of coverage gaps rather than silent data loss. And under the Team mechanism, data sovereignty, since telemetry remains in the customer’s own S3 bucket.
The API mechanism for Claude Enterprise is available today, with the OpenTelemetry mechanism for Claude Team following.
Source: Businesswire