Getting Started

Quickstart

Route your first agent request through the Spineforge gateway in about five minutes, no SDK, no code changes.

How it works

Spineforge acts as a drop-in reverse proxy. Your agent keeps using the OpenAI-compatible API it already speaks; you only change where it sends requests and which key it presents.

1

Base URL: point your agent at the Spineforge gateway

https://spineforge-backend.onrender.com/gateway/v1
2

API key: use your Spineforge gateway key

sfgw-xxxx-xxxx

1. Add your provider key to the Vault

Sign in to the Spineforge dashboard and add your actual LLM provider key (for example your Google or OpenAI key) to the Dashboard Vault. The gateway resolves it securely at request time, so agents never see it.

2. Create a gateway key

In the dashboard, open the API Keys tab and create a gateway key for your agent. Gateway keys start with sfgw-. Create one key per agent (or per environment) so every request can be attributed, budgeted and revoked independently. You can also set a budget cap, a rate limit and the provider / model scopes the key is allowed to use.

3. Change the Base URL and the key

Point your agent at the gateway URL and use your gateway key as the API key. Spineforge supports any tool that lets you override an OpenAI-compatible base URL. See the full guides for IDEs, CLIs and custom & cloud-hosted agents.

Connecting IDEs & Tools

  • Cursor: Go to Settings > Models > OpenAI API Key. Enter your Gateway Key (sfgw-...). Click "Override OpenAI Base URL" and enter https://spineforge-backend.onrender.com/gateway/v1.
  • OpenCode: Run opencode auth login, select "Other Custom provider", and provide your Gateway Key and Base URL when prompted.
  • VS Code (Native Chat): Run Chat: Manage Language Models, then Add Models > Custom Endpoint, and add a model that points at the gateway. Full settings.json block in the IDE guide.
  • Antigravity CLI: Export GOOGLE_GEMINI_BASE_URL=https://spineforge-backend.onrender.com/gateway/v1 and set your Gateway Key in your config.
  • VS Code Extensions: For third-party OpenAI-compatible extensions (like Continue), set the provider to "OpenAI", paste your Gateway Key, and set the API Base URL to the Gateway URL.
  • Other CLIs & IDEs: The universal logic is simple: select "OpenAI" (or "OpenAI Compatible") as the provider, paste the Gateway URL as the Base URL, use your Gateway Key as the API key, and define your model.
Important prerequisite: you must add your actual LLM provider key (e.g., your Google or OpenAI key) to the Spineforge Dashboard Vault so the Gateway can securely resolve it.

Or test with a simple script

If you prefer to test with a script instead of an IDE, here is a 5-line example using the standard OpenAI SDK:

from openai import OpenAI

client = OpenAI(
    base_url="https://spineforge-backend.onrender.com/gateway/v1",
    api_key="sfgw-xxxx-xxxx"  # Your Gateway Key
)

# Your real API keys are securely injected by the Spineforge Vault!
print(client.chat.completions.create(model="openai/gpt-4o", messages=[{"role": "user", "content": "Hi!"}]))

Or with cURL:

curl https://spineforge-backend.onrender.com/gateway/v1/chat/completions \
  -H "Authorization: Bearer sfgw-xxxx-xxxx" \
  -H "Content-Type: application/json" \
  -d '{"model": "openai/gpt-4o", "messages": [{"role": "user", "content": "Hi!"}]}'
Model names: names that contain a / need a provider prefix such as openai/gpt-4o or gemini/gemini-3.5-flash. See Model Naming.

4. See it in the dashboard

Open the dashboard and your request appears with its agent, provider, model, tokens, latency and cost. From there you can watch spend, view the agent on the graph, and use the kill-switch to revoke the key instantly.

  • Real-time cost and usage per agent, provider and model
  • Per-agent timeline and lifecycle status
  • Budget caps, rate limits, and Governance & Scopes controls

Next steps