Integrate your agent

Rungate is a drop-in replacement for the OpenAI and Anthropic API endpoints. Point your existing SDK at Rungate's base URL, swap your provider key for an rg_agt_* token, and the request/response shape is unchanged.

Base URL: https://app.rungate.dev/v1

OpenAI SDK · Python

from openai import OpenAI

client = OpenAI(
    base_url="https://app.rungate.dev/v1",
    api_key="rg_agt_your_token_here",
)

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "..."}],
    extra_headers={
        "x-rungate-run-id": "run_customer_refund_2026_04_26",
        "x-rungate-policy": "prod-agents/v4",
    },
)

# Streaming
stream = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "..."}],
    stream=True,
    extra_headers={"x-rungate-run-id": "run_customer_refund_2026_04_26"},
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")

OpenAI SDK · TypeScript

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://app.rungate.dev/v1',
  apiKey: 'rg_agt_your_token_here',
});

const response = await client.chat.completions.create({
  model: 'gpt-4',
  messages: [{ role: 'user', content: '...' }],
}, {
  headers: {
    'x-rungate-run-id': 'run_customer_refund_2026_04_26',
    'x-rungate-policy': 'prod-agents/v4',
  },
});

Anthropic SDK · Python

from anthropic import Anthropic

client = Anthropic(
    base_url="https://app.rungate.dev/v1",
    api_key="rg_agt_your_token_here",
)

message = client.messages.create(
    model="claude-sonnet-4",
    max_tokens=1024,
    messages=[{"role": "user", "content": "..."}],
    extra_headers={
        "x-rungate-run-id": "run_customer_refund_2026_04_26",
    },
)

Anthropic SDK · TypeScript

import Anthropic from '@anthropic-ai/sdk';

const client = new Anthropic({
  baseURL: 'https://app.rungate.dev/v1',
  apiKey: 'rg_agt_your_token_here',
});

const message = await client.messages.create({
  model: 'claude-sonnet-4',
  max_tokens: 1024,
  messages: [{ role: 'user', content: '...' }],
}, {
  headers: {
    'x-rungate-run-id': 'run_customer_refund_2026_04_26',
  },
});

curl

curl -X POST https://app.rungate.dev/v1/chat/completions \
  -H "Authorization: Bearer rg_agt_your_token_here" \
  -H "x-rungate-run-id: run_customer_refund_2026_04_26" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4",
    "messages": [{"role": "user", "content": "Hello, Rungate."}]
  }'

Headers reference

All directives work as x-rungate-* headers, or as body fields under a rungate key (use the body form if your SDK strips unknown headers):

Header	Body field	Purpose
x-rungate-run-id	rungate.run_id	Stable ID grouping calls into one run.
x-rungate-new-run	rungate.new_run	Boolean. Force a new run on this call.
x-rungate-policy	rungate.policy	Override the agent's default policy.
x-rungate-user	rungate.user	End-user identifier for analytics.
x-rungate-tags	rungate.tags	Comma-separated tags.

Frameworks

Rungate is provider-shape compatible, so any framework that lets you configure a custom base_url works out of the box. A few popular ones:

LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://app.rungate.dev/v1",
    api_key="rg_agt_your_token_here",
    model="gpt-4",
    default_headers={"x-rungate-run-id": "run_..."},
)

LlamaIndex

from llama_index.llms.openai import OpenAI

llm = OpenAI(
    api_base="https://app.rungate.dev/v1",
    api_key="rg_agt_your_token_here",
    model="gpt-4",
    additional_kwargs={"extra_headers": {"x-rungate-run-id": "run_..."}},
)

OpenClaw

Set the RUNGATE_BASE_URL and RUNGATE_AGENT_KEY environment variables. OpenClaw routes provider calls through Rungate automatically.

export RUNGATE_BASE_URL=https://app.rungate.dev
export RUNGATE_AGENT_KEY=rg_agt_your_token_here

What Rungate forwards unchanged

• The full request body (any OpenAI Chat Completions or Anthropic Messages payload).
• Streaming (stream: true). The response is a standard SSE stream.
• Function calls / tool use — the provider's native shape.
• Vision / multimodal payloads.

What Rungate may add or change

• Governance response codes (202, 402, 403, 429) before forwarding to the provider.
• Cross-provider failover — the response format the agent sent is preserved, but the actual response may come from a fallback provider.
• An X-Rungate-Run-Id response header echoing the run the call was attached to.

Failure modes

See response codes in the agent reference for the full table. Quick map:

• 200 — success.
• 202 — approval gate (see Approval gates).
• 402 — budget exceeded (see Policies & budgets).
• 403 — policy violation.
• 409 — run already closed; start a new run.
• 429 — rate limit; honor Retry-After.
• 5xx — Rungate or upstream provider error; exponential backoff.