Anthropic API
Agents built on the native Anthropic Messages API (Claude Code, the Claude Agent SDK, or the @anthropic-ai/sdk client) can’t use the OpenAI-compatible surface. Use the gateway’s /anthropic passthrough endpoint instead. This is the only way to run a Claude-Code-based agent on the gateway.
The platform injects only ASTRO_GATEWAY_URL and ASTRO_GATEWAY_API_KEY. It does not set any ANTHROPIC_* variables. You map them yourself, and there are three gateway-specific details:
Authenticate with the x-bf-vk header, not a Bearer token. The gateway ignores Authorization and x-api-key. For Claude Code, also set CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1 to disable its pre-release beta flags, which Bedrock rejects.
Streaming, client-defined tool use, and adaptive thinking all work over this path. Anthropic’s server-side tools (web_search, web_fetch, code execution, computer use) are not available on the Claude models, because they run on Amazon Bedrock, which doesn’t host them. For web search, use one of the gpt-* models on the OpenAI-compatible surface (see Web search). Otherwise wire up your own tool and feed the result back as a tool result.
Next steps
- Monitor Claude Code: send Claude Code telemetry to your dashboard
- Errors and limits: status codes and gateway boundaries