Anthropic is an AI safety and research platform that develops the Claude family of large language models (LLMs). While Claude models can be accessed across various cloud infrastructure platforms, the Anthropic AI integration specifically connects to the native Anthropic API platform. When you configure the Anthropic AI integration, Collibra ingests the metadata of Claude models and Claude Managed Agents hosted directly within your Anthropic developer account.
For information about Anthropic, go to the Anthropic documentation.
Key capabilities
The Anthropic AI integration helps you:
- Ingest model metadata: Automatically catalog all Claude model versions available through your Anthropic API account.
- Govern AI agents: Govern Claude Managed Agents as distinct AI Agents in Collibra, including tracking their platform-hosted configurations and system prompts.
- Trace AI lineage: Establish clear relationships between your active platform-hosted agents, model versions, specific model deployments, and API endpoints.
-
You can configure Edge connections and capabilities without an active AI Governance license. However, AI Governance must be enabled to harvest AI model metadata, ingest corresponding AI assets in Data Catalog, and access the dashboards and features necessary to visualize and govern your AI landscape.
- You can integrate Anthropic AI only via Edge.
- To use the Anthropic AI integration, you need Collibra Platform 2026.07 or newer.
- A separate Edge connection is required for each Anthropic workspace. Anthropic standard API keys are scoped to a single workspace, so one connection cannot span multiple workspaces.
Integration scope
To govern the right data, note the distinction between the platform-managed metadata that the Anthropic AI integration catalogs and the local execution environments that it does not.
| Category | In scope | Out of scope |
|---|---|---|
| Models | Official Claude models: Standard versions and metadata available via the Anthropic API. |
Third-Party clouds: Claude models deployed via Amazon Bedrock or Gemini Enterprise Agent Platform. |
| Agents | Claude Managed Agents: Agent configurations, parameters, and system prompts explicitly hosted and managed on the Anthropic platform infrastructure. |
Claude Code and local agents: The Claude Code CLI, Claude Agent SDK, custom LangChain/LlamaIndex setups, or any agent loops executed locally on developer laptops. |
| Endpoints |
Anthropic API endpoints: The connection points and usage metrics managed by your Anthropic developer account. |
Workplace applications: Internal or enterprise applications that call the Anthropic SDK but handle agent orchestration logic client-side. |