OpenAI-compatible API Gateways Integration Guide

Overview

OpenAI-compatible API Gateways are governed by routing traffic through a managed proxy to attribute outbound activity, apply policy, and shield API keys.

Support level

Compatibility Adapter + Provider/API Proxy

What Cup’n’String detects

• Custom base URLs
• Known provider domains
• API-key-bearing environment variables/config files
• Local proxy endpoints

What it governs

• Approved provider access
• API key shielding
• Model gateway routing
• Audit of agent/provider traffic

Recommended policies

• Require provider allowlist
• Use organization-approved proxy URLs
• Block direct calls to unapproved providers
• Redact API keys from logs and prompts

Setup outline

1. Ensure the Cup’n’String client is running.
2. Configure your agent or IDE to use the local Cup’n’String gateway URL as the base URL.
3. Configure the agent daemon with the target gateway endpoint details and API keys.

Verification

Send a model request to the gateway address and verify it is captured and logged.

Troubleshooting

If requests fail, check that the local proxy port matches the configured base URL in your IDE configuration.

Known limitations

Provider-specific metadata varies across APIs and compatibility layers.