← Back to tools

Claude

Anthropic · Assistant

Approved Cleared up to Confidential

Anthropic's assistant (Claude.ai + Claude Code). Approved up to Confidential under the enterprise agreement.

Policy note: Enterprise/Team plans covered by the signed agreement. MCP connections are reviewed per-server.

Platforms

2 of 3 approved

Where Claude can be accessed from — each route reviewed independently.

Direct vendor

Approved

AWS

Approved

Google Cloud

Under review

Connectivity map

Approved Conditional Under review Denied

How Claude operates and what it can reach — each path approved independently.

Work modes

3 of 4 approved

Each way of operating Claude is reviewed independently.

Chat

Standard conversation in Claude.ai.

Approved

Projects

Persistent knowledge + instructions scoped to a workspace.

Approved

Claude Code

Agentic coding in the terminal/IDE.

Approved

Computer Use

Controls a desktop by taking screenshots and actions.

Denied pending endpoint-control review.

Denied

Connections

3 of 5 approved

MCP servers and integrations Claude can reach, each reviewed independently.

GitHub MCP MCP

MCP server for repository read/write.

Approved

Filesystem MCP MCP

Local filesystem access via MCP.

Approved for non-Restricted local repos only.

Conditional

Atlassian MCP MCP

Jira + Confluence read/write via MCP.

Approved

Web Search Built-in

Built-in web search.

Approved

Slack Connector Connector

Reads/sends Slack messages.

Under review