Configuration Export

What this page does

Configuration export turns four choices into a copyable or importable tool configuration:

• An enabled API key from Agent API Keys.
• A model that is available to that key and target client.
• The correct Base URL for standard, image, or video traffic.
• The tool format, such as Claude Code, Cursor, Python SDK, cURL, or CC Switch.

Use Quickstart for a single manual request. Use this page when the same key, model, and route need to be reused across local tools or a team workflow.

Recommended setup flow

1. Create or select an API key in Agent API Keys
2. Open Configuration export
3. Select the key, model, and Base URL
4. Choose a target tool tab
5. Copy the generated config or import it into CC Switch
6. Send one request from the target tool and verify the response

1. Create or select an API key in Agent API Keys
2. Open Configuration export
3. Select the key, model, and Base URL
4. Choose a target tool tab
5. Copy the generated config or import it into CC Switch
6. Send one request from the target tool and verify the response

Base URL rules

Standard API tools and SDKs
  Base URL: https://api.gpt88.cc
  Use the endpoint path, headers, and body fields required by the API.

Direct image and video APIs
  Base URL: https://img.gpt88.cc
  Follow the image or video API reference for /v1 or /v1beta paths.

All tools
  Do not append a second /v1 manually.

Keep the Base URL at the protocol boundary. The SDK or tool should append its resource path. For example, a chat client normally uses https://api.gpt88.cc/v1, while direct image and video workflows usehttps://img.gpt88.cc according to their API reference.

Supported tool patterns

• Claude Code: use the generated Claude-compatible values and keep the API key in the process environment.
• Cursor: select an OpenAI-compatible provider, then verify the Base URL and exact model ID.
• Python or Node.js: export the key and Base URL as server-side environment variables.
• cURL: use a short request as the fastest connectivity check before enabling tools.
• CC Switch: use the generated import target when the installed version supports it; otherwise copy the import link manually.

Security and troubleshooting

• Never commit an exported configuration, screenshot, or full API key to a public repository or chat.
• Store keys in environment variables or a secret manager. Disable and rotate a key immediately if it may have leaked.
• 401: check the Bearer header and confirm that the key is enabled.
• 404: check the host, the single /v1 prefix, and the exact model ID.
• 429: inspect usage and reduce concurrency before adding bounded exponential backoff.
• Timeout: validate DNS, TLS, proxy settings, and the shortest request with cURL first.

Verify the generated configuration

Before enabling a long-running agent workflow, send one minimal request and save the result metadata:

curl https://api.gpt88.cc/v1/chat/completions \
  -H "Authorization: Bearer $GPT88_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.6-sol","messages":[{"role":"user","content":"Reply with OK"}],"max_tokens":32}'

Confirm the HTTP status, returned model, request ID, and the usage record in the console. Only then add long context, streaming, tool calls, or automated retries.