Integration

Endpoints

Send the key as Authorization: Bearer sk-…

Claude Code

Claude Code reads these two variables at startup. Add them to your shell profile to make them persist across sessions.

export ANTHROPIC_BASE_URL="https://api.shuntio.com"
export ANTHROPIC_AUTH_TOKEN="sk-your-key"

Use ANTHROPIC_AUTH_TOKEN, not ANTHROPIC_API_KEY — the latter is validated against Anthropic and rejected.

Codex CLI

Add a provider to ~/.codex/config.toml, then run Codex with that profile.

[model_providers.shuntio]
name = "shuntio"
base_url = "https://api.shuntio.com/v1"
env_key = "SHUNTIO_API_KEY"

[profiles.shuntio]
model_provider = "shuntio"
model = "your-model-id"

export SHUNTIO_API_KEY="sk-your-key"
codex --profile shuntio

Keep the profile name and the provider name distinct; they are separate keys in the file.

OpenAI SDK

The only change is base_url. Request and response shapes are untouched.

Python

from openai import OpenAI

client = OpenAI(
    base_url="https://api.shuntio.com/v1",
    api_key="sk-your-key",
)

Node.js

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.shuntio.com/v1',
  apiKey: process.env.SHUNTIO_API_KEY,
});

curl

curl https://api.shuntio.com/v1/chat/completions \
  -H "Authorization: Bearer $SHUNTIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "your-model-id",
    "messages": [{ "role": "user", "content": "ping" }]
  }'

Cursor, Cherry Studio, and other clients

Any client that accepts a custom OpenAI-compatible endpoint needs the same three values.

Verify

This should list the models your key can reach.

curl -s https://api.shuntio.com/v1/models \
  -H "Authorization: Bearer $SHUNTIO_API_KEY"

If something fails

Start from the status code:

• 401 — the key is missing, mistyped, or revoked. Check that the header isAuthorization: Bearer sk-…
• anything else: the console request log records the exact upstream response for that call, which is more reliable than guessing from the status.