Trust3 AI adds security visibility to Anthropic Claude Code

Closing the visibility gap

Trust3 AI has launched an integration between its Agent DOS platform and Anthropic’s Claude Code. This move is designed to give enterprise security and AI governance teams a clear view of how autonomous coding agents behave within a company. Because Claude Code runs directly on developer machines, it can read and modify source code while calling various tools and Model Context Protocol (MCP) servers.

Before this integration, security teams lacked a reliable way to see who was using Claude Code or what systems it was connecting to. Organizations often resisted using endpoint agents—software that monitors every action on a machine—because of the security risks and the friction they cause for developers. Trust3 AI aims to solve this by providing oversight without requiring those intrusive software installations.

The goal is to ensure that code written by AI agents is held to the same standards as human-written code. As Balaji Ganesan, CEO and Co-founder of Trust3 AI, stated, "Agent DOS gives security complete oversight, so agent-written code ships under the same scrutiny." This allows companies to embrace faster development cycles without sacrificing their internal governance policies.

Two pathways for data

The integration works differently depending on whether a company uses Claude Enterprise or Claude Team. For Claude Enterprise customers, the system uses an API pull mechanism. This method leverages Anthropic’s Compliance and Analytics APIs to gather session transcripts, tool calls, and per-user usage data.

Claude Team customers use a different approach involving an admin-pushed OpenTelemetry configuration. This setup directs Claude Code to send telemetry to the customer’s own OpenTelemetry Collector and S3 bucket. The Trust3 AI Collector then reads that data, ensuring that the telemetry remains within the customer's own storage environment for better data sovereignty.

Both of these pathways are described as agentless and read-only. They feed into the same Collector, Governance, and Agent DOS pipeline that Trust3 AI already uses for other platforms. This consistency allows the company to extend its existing agent, identity, and trace models to include autonomous coding agents.

Accountability and inventory

One of the primary benefits of this integration is the ability to create a complete inventory of users. Every Claude Code user is treated as an accountable AI agent and is assigned a named owner. This helps security teams understand exactly who is responsible for the actions taken by the AI during a coding session.

The system also provides visibility into specific technical interactions, such as MCP server connections and tool calls. These details are pulled from whichever telemetry source the organization's specific plan provides. This level of detail is critical for identifying which internal systems are being accessed by the AI and what specific commands are being executed.

Furthermore, the platform is designed to be transparent about what it cannot see. If telemetry fails to reach the collector for any reason, Agent DOS reports the coverage gap explicitly. This prevents the system from dropping data silently, which ensures that security teams are always aware of the limits of their current visibility.

Enterprise-wide coverage

Trust3 AI is expanding its reach by bringing these capabilities to multiple major platforms. The Agent DOS platform is already being used to manage governance for Copilot, Foundry, M365, and Databricks. By adding Claude Code, the company is positioning itself as a central hub for managing various types of AI agents.

The integration is designed to scale across large organizations that have complex security requirements. By providing a way to monitor Claude Code without touching the developer's local environment, it reduces the barrier to entry for many IT departments. It allows for a smoother transition into using autonomous agents for production-level software development.

The rollout of these features is happening in stages. The API mechanism for Claude Enterprise customers is available immediately. The OpenTelemetry mechanism for Claude Team customers is scheduled to follow shortly after the initial release.

Impact on development

For developers, this means they can continue using powerful tools like Claude Code without being hindered by heavy monitoring software. The focus remains on productivity while the underlying governance happens through the read-only pathways provided by Trust3 AI. This balance helps maintain a high velocity for engineering teams while satisfying corporate compliance.

For security leaders, the integration provides a standardized way to audit AI behavior. Instead of guessing what an autonomous agent might do, they can see the actual transcripts and tool calls. This data is essential for identifying potential risks before code is deployed to production environments.

As more companies adopt autonomous coding agents, the need for these types of governance platforms will likely grow. Trust3 AI's approach of using existing APIs and OpenTelemetry standards suggests a move toward a more integrated and less intrusive way of managing AI in the workplace.

Technical details

The integration relies on the Model Context Protocol (MCP), which is a standard for how AI models connect to data sources and tools. By monitoring these connections, Trust3 AI can see exactly what resources the AI is interacting with. This is a key component of the trace model they use to track agent activity.

The use of OpenTelemetry is also significant for the Claude Team pathway. OpenTelemetry is a framework for collecting logs, metrics, and traces from software. By using it, Trust3 AI allows customers to keep their data in their own S3 buckets, which is a major requirement for many companies with strict data residency rules.

The platform ensures that all data flows into a unified pipeline. Whether the data comes from an API pull or an OpenTelemetry collector, it ends up in the same Governance and Agent DOS pipeline. This unified approach makes it easier for administrators to manage security policies across different types of AI tools simultaneously.

Next steps for teams

Organizations already using Claude Enterprise can begin utilizing the API pull mechanism today. This is the fastest way to gain visibility into Claude Code activity for existing high-level plans. Security teams can start by auditing current usage and identifying the most active AI agents.

Teams using Claude Team should prepare for the upcoming OpenTelemetry configuration. This will require an admin-pushed configuration to route telemetry to their own S3 buckets. It is a good time for these teams to ensure their storage buckets and collectors are ready to receive the incoming data.

As these tools become more common, companies should look for ways to standardize their AI agent governance. Using a platform that can handle multiple tools like Copilot and Claude Code in one place can simplify the task of maintaining a secure and transparent AI environment.

Summary of features

The Trust3 AI integration provides four specific capabilities for technical AI and security leaders. First, it creates a complete inventory of every Claude Code user as an accountable AI agent with a named owner. Second, it provides visibility into MCP server connections and tool calls based on the organization's plan.

Third, the platform explicitly surfaces coverage gaps rather than allowing data to be lost silently. This ensures that administrators know exactly which parts of their environment are being monitored. Fourth, the Team mechanism ensures data sovereignty by keeping telemetry within the customer's own S3 bucket.

By combining these features, Trust3 AI offers a comprehensive security layer for Claude Code. It addresses the needs of both security teams who require oversight and developers who require a frictionless experience. The integration marks a significant step in making autonomous coding agents safer for enterprise use.

Implementation details

The integration is designed to be non-intrusive for the end user. Because it is agentless, it does not require the installation of any software on the developer's machine. This is a major selling point for organizations that are wary of the security implications of installing third-party agents on every workstation.

The read-only nature of both pathways ensures that the security platform cannot modify the data it collects. It simply observes the activity of Claude Code as it interacts with the system. This protects the integrity of the source code while still providing the necessary logs for compliance and auditing.

The platform's ability to handle different types of telemetry sources allows it to be flexible. It can adapt to the specific needs of different customer tiers while maintaining a consistent backend for governance. This scalability is key for enterprises that may use a mix of different AI tools across various departments.

Governance and compliance

Enterprise governance often requires a clear audit trail for every action taken in a production environment. When AI agents are involved, this audit trail must include the agent's actions, the tools it used, and the human owner who authorized its use. Trust3 AI provides this by linking every session to a named owner.

The inclusion of session transcripts and tool calls provides a deep level of detail. Security teams can review these transcripts to ensure the AI is following company policies and not performing unauthorized actions. This is especially important when the AI is granted the ability to modify source code or interact with internal servers.

By providing a clear view of these interactions, Trust3 AI helps companies meet their compliance obligations. It allows them to demonstrate that they have control over their AI tools. This is a necessary step for many industries that are subject to strict regulations regarding software development and data security.

Future of AI agents

The move toward autonomous coding agents is a major trend in software development. Tools like Claude Code allow for faster iteration and the automation of repetitive coding tasks. However, as these tools become more powerful, the need for robust security and governance frameworks becomes more urgent.

Trust3 AI's integration with Claude Code is part of a broader effort to secure the AI ecosystem. By providing a platform that can manage agents across multiple services, they are creating a centralized way to handle AI governance. This will likely become a standard requirement for enterprises as they scale their use of AI.

As more organizations adopt these technologies, we can expect to see more integrations like this one. The focus will remain on providing visibility and control without creating friction for the people doing the work. Trust3 AI's approach of using existing APIs and standards is a key part of that strategy.

Data sovereignty and privacy

Data sovereignty is a major concern for many large enterprises, especially those in regulated industries. They need to ensure that their sensitive data does not leave their controlled environment. The OpenTelemetry mechanism for Claude Team customers directly addresses this concern by keeping telemetry in the customer's own S3 bucket.

This approach allows companies to maintain full control over their data while still benefiting from the insights provided by Trust3 AI. The platform reads the data from the customer's bucket to provide governance, but the data remains under the customer's ownership. This is a critical distinction for many IT and security departments.

By offering these different pathways, Trust3 AI can serve a wide range of customers. They can provide a simpler API-based solution for some, while offering a more robust, sovereign solution for others. This flexibility is essential for a platform that aims to be the standard for enterprise AI agent security.

Technical architecture

The technical architecture of the integration is built on standard protocols. By using Anthropic’s Compliance and Analytics APIs, Trust3 AI can pull data directly from the source for Enterprise users. This avoids the need for any intermediate software or complex configurations for those customers.

For Team users, the use of OpenTelemetry is a standard way to collect telemetry from applications. By directing Claude Code to send its telemetry to a customer-owned collector, the system follows a standard architectural pattern. This makes it easier for existing IT infrastructure to incorporate the new data stream.

The final step in the pipeline is the Trust3 AI Collector, which processes the data from both pathways. This unified backend ensures that the governance rules and reporting are consistent regardless of how the data was collected. This architecture provides a scalable and reliable way to monitor AI activity across an entire organization.

Deployment and rollout

The rollout of the Trust3 AI and Claude Code integration is designed to be phased. This allows customers to adopt the new features at their own pace. The immediate availability of the API mechanism for Claude Enterprise customers allows for a quick start for those who need it most.

The subsequent release of the OpenTelemetry mechanism for Claude Team customers ensures that all types of organizations will eventually have access to the same level of security. This phased approach helps manage the complexity of the rollout while ensuring that all users eventually benefit from the new capabilities.

As more organizations adopt these tools, the feedback from these early users will help refine the platform. Trust3 AI's goal is to provide a seamless experience that enhances security without hindering the benefits of AI-driven development. The ongoing updates to the platform will reflect this commitment to both security and usability.

Impact on security teams

Security teams are often the gatekeepers for new technology in the enterprise. They need to be convinced that a new tool is safe before they allow it to be used. Trust3 AI's platform provides the evidence they need to approve Claude Code by giving them full visibility into its actions.

With a complete inventory of AI agents and a clear view of their activities, security teams can set and monitor policies more effectively. They can see which tools are being accessed and by whom, allowing them to identify and mitigate risks quickly. This proactive approach is much more effective than trying to react to security incidents after they occur.

The ability to see coverage gaps also helps security teams identify areas where they might need to improve their monitoring. This continuous feedback loop allows them to strengthen their security posture over time. By providing these tools, Trust3 AI helps security teams become enablers of AI adoption rather than roadblocks.

Developer experience

One of the biggest challenges in enterprise security is maintaining a good developer experience. Developers often find security tools to be intrusive or slow, which can lead to them finding ways to bypass those tools. Trust3 AI's agentless approach is designed to avoid this problem entirely.

Because there is no software to install on the developer's machine, their workflow remains uninterrupted. They can use Claude Code to write and modify code as quickly as they can, while the security and governance teams get the data they need in the background. This

Summary of integration

The integration between Trust3 AI and Anthropic’s Claude Code is a significant step forward for enterprise AI security. It provides a way for companies to use autonomous coding agents while maintaining full oversight of their activities. By offering two different pathways for data collection, it caters to the needs of both Claude Enterprise and Claude Team users.

The platform provides a complete inventory of AI agents, visibility into tool calls, and explicit reporting of coverage gaps. It also ensures data sovereignty for customers who require it. These features combined make it a powerful tool for any organization looking to adopt AI-driven development safely.

As the technology matures, this type of integration will likely become a standard part of the AI development lifecycle. It allows for a balance between the speed of AI and the security requirements of the enterprise. Trust3 AI is positioning itself as a leader in this space by providing a clear, scalable, and non-intrusive solution for AI governance.

Final thoughts

The launch of this integration highlights the growing importance of AI governance in the enterprise. As companies move from experimenting with AI to deploying it in production, the need for visibility and accountability becomes paramount. Trust3 AI's platform addresses these needs directly by providing a clear audit trail for AI actions.

By focusing on an agentless, read-only approach, Trust3 AI minimizes the friction of security for developers. This is a crucial factor in the successful adoption of new technologies. When security is easy to implement and doesn't get in the way of work, it is much more likely to be embraced by the entire organization.

For security leaders and AI governance teams, this integration offers a way to manage the risks of autonomous coding agents without sacrificing the benefits. It provides the oversight needed to ensure that AI-written code meets the same standards as human-written code. This is a vital step toward a more secure and productive future for software development.

The role of MCP

The Model Context Protocol (MCP) is a key part of how Claude Code interacts with the outside world. It allows the AI to connect to various data sources and tools, such as databases, file systems, and internal APIs. Because these connections can be a source of risk, monitoring them is essential for security teams.

Trust3 AI's integration provides visibility into these MCP server connections and the specific tool calls made during a session. This allows security teams to see exactly what resources the AI is accessing and what actions it is performing. This level of detail is necessary for identifying potential unauthorized access or data leaks.

By providing this visibility, Trust3 AI helps companies understand the full scope of their AI's capabilities and limitations.