https://mcp.1eye.ai/mcp. A production key begins with 1eye_live_. Use the Bearer header shown here for MCP connections.
Create and manage a key
Open Developers → API Keys in 1eye. Use a descriptive name for each integration so you can identify the key to rotate or revoke later. Key creation and management require the appropriate workspace access; see API Keys. The full secret is shown once at creation. Store it in your client’s supported secret mechanism or a secrets manager. Environment variable references can keep the literal key out of a shared configuration file. Never include a key in a URL, public repository, screenshot, prompt, or published example. When sharing diagnostics, redact theAuthorization header and any environment variable value containing the key.
Workspace scope
Each call operates on the workspace associated with the API key. The MCP tools do not accept a workspace ID to switch workspaces. To use another workspace, configure the client with a key from that workspace. Verify the selection withget_workspace before running reports or approving changes.
The available MCP catalog describes which operations exist. It does not, by itself, establish a separate read-only API-key scope or a complete role permission matrix. Use your workspace’s access controls and your client’s tool restrictions as applicable.
Read tools and changes
The current catalog has 19 read tools and 11 tools that change records or settings. Each change tool’s description instructs the assistant to obtain confirmation first.
Configure the client to ask before writes or restrict the tools available to an agent. The instruction to ask for confirmation depends on client behavior; it is not a server-side authorization guarantee.
For a reporting-only integration, allow only the read tools listed in the tool reference. A client tool restriction controls that client, not the underlying key’s access in other integrations.