AnyInt Docs

Connect Agent Tools to AnyInt

Choose an agent tool, add your AnyInt API key and model ID, then verify the connection with one command.

Connect an existing agent product to AnyInt, or use AnyInt APIs to build your own agent application. Choose the path that matches what you are trying to finish.

Before you start

  1. Create an AnyInt API key.
  2. Check the models available to your account, or call the Models API.
  3. Keep the key in an environment variable or secret manager. Do not commit it to a repository.
export ANYINT_API_KEY="your-anyint-api-key"

curl https://api.anyint.ai/openai/v1/models \
  -H "Authorization: Bearer $ANYINT_API_KEY"

Use a model ID returned for your account. Do not assume that every public model name is enabled for every key.

Choose your tool

Support status

Protocol status and tool ownership are separate:

ToolAnyInt API familyProtocol statusIntegration ownership
Claude CodeAnthropic-compatibleStableAnyInt setup guide for a third-party client
CodexOpenAI ResponsesBetaAnyInt setup guide for a third-party client
Gemini CLIGemini-compatibleStableAnyInt setup guide for a third-party client
CC-SwitchDepends on the selected clientInherits the API family statusThird-party provider manager
Other toolsOpenAI- or Anthropic-compatibleDepends on the selected familyCompatibility guide

Beta routes should be verified with the target account, model, and client version before production use.

Build an agent application

If you control the application code, start with the API family your SDK already supports:

NeedStart here
OpenAI SDK or Chat CompletionsOpenAI-compatible API
OpenAI Responses-based agentsOpenAI Responses API
Anthropic Messages or Claude-style content blocksAnthropic-compatible API
Gemini-native content and function declarationsGemini-compatible API
Functions or external actionsTool Calling
Predictable machine-readable outputStructured Outputs
Incremental outputStreaming

Completion checklist

Your setup is complete when:

  • the client uses an AnyInt public base URL
  • the API key is not stored in source control
  • the configured model ID is visible to the same key
  • a minimal prompt returns a response
  • the client can recover from an invalid key, invalid model, and temporary upstream error

Continue with Troubleshooting if the first request does not complete.

On this page