Overview
Use these recipes to point coding agents and IDE tools at ApiGo’s unified gateway. You keep one sk-apigo-... key, pick model IDs from the Model Catalog, and switch tools without juggling vendor credentials.
Before you start
- Create an API key in Workspace → API Keys.
- Confirm the base URL for your client:
| Protocol | Base URL | Typical clients |
|---|---|---|
| OpenAI-compatible | https://api.apigo.ai/v1 |
Codex CLI, Cursor, DeepSeek Harness, Kimi Code, OpenCode, Hermes, OpenClaw (OpenAI path) |
| Anthropic Messages | https://api.apigo.ai |
Claude Code, Claude Desktop, OpenClaw (Anthropic path) |
- Copy the exact model slug from the catalog. For example, use
gpt-5.6-solwith the OpenAI-compatible path orclaude-sonnet-5with Anthropic Messages.
Coding agents
| Guide | Tool |
|---|---|
| Claude Code | Anthropic’s terminal coding agent |
| Claude Desktop | Claude Desktop third-party Gateway |
| Codex CLI | OpenAI’s local coding agent |
| Cursor | Cursor BYOK / OpenAI base URL override |
| DeepSeek Harness | DeepSeek AI’s open-source agent harness |
| Hermes Agent | Nous Research terminal agent |
| Kimi Code | Moonshot AI’s terminal coding agent |
| OpenClaw | Multi-channel agent platform |
| OpenCode | Open-source TUI / desktop coding agent |
Why route agents through ApiGo
- One key for many upstream models
- Workspace spend caps, RPM limits, and routing policies
- Shared usage and call logs across CLI, IDE, and backend traffic
After setup, send a short prompt in the agent and confirm the request in Workspace → Call Logs.
