FAQ
Common questions, answered precisely.
Do I have to change my agent code to use Spineforge?
No code changes are needed: just change the Base URL and API key in your app, IDE or CLI. Spineforge is a drop-in reverse proxy for the OpenAI-compatible API.
What do I need to integrate an agent?
Two things: the gateway URL as the base URL, and a Spineforge gateway key (sfgw-...) as the API key. Your real provider keys are added once to the Dashboard Vault.
Which tools and agents are supported?
Anything that lets you set an OpenAI-compatible base URL and API key: IDEs such as Cursor and VS Code, CLIs such as OpenCode and Antigravity CLI, custom agents in Python or Node.js (including LangChain), visual builders such as Flowise and Langflow, and cloud-hosted agents in containers, Kubernetes, serverless or CI. Antigravity IDE is not supported yet.
Does Spineforge add latency to my agent calls?
Yes, a small amount. Because the gateway sits in the data path between your agent and the LLM, it adds a minimal network hop: roughly 15–35 ms. The proxy layer is stateless and horizontally scaled, and streamed responses are piped straight through.
What happens if the Spineforge gateway is unreachable?
Your agent will receive connection errors and cannot reach the LLM through the gateway until it is back online. We target 99.9% uptime; see the Service Reliability page for measured numbers.
Is revocation an instant kill-switch?
Yes. Because traffic flows through Spineforge, revoking a gateway key instantly blocks all in-flight and future requests (403 Forbidden). The agent never held a provider key, so there is nothing cached to expire.
Where do my real provider keys live?
Only in the Spineforge Vault, encrypted and isolated. Agents and developers only ever hold gateway keys, so your root LLM credentials never appear in local environments, IDEs or client-side code.
Why is there a migration requirement for credential custody?
Credential custody only works if raw provider keys exist exclusively in Spineforge. If a key also lives in an env file or shared vault, any process that can read it can use the key, registered or not. Removing keys from other locations is not optional fine print; it's the step that makes the security guarantee hold.
Why is cost stored and not recomputed, but latency is live?
Cost is a fact about a past event: it happened at a specific rate, and that rate should be frozen. A request that cost $0.03 three months ago should still show $0.03 today, even if rates have changed since. Latency, by contrast, is a property of a window and a dataset: adding a new request today meaningfully changes the P95. These have different semantics, so we treat them differently.
How do I trace a request's cost back to the rate that produced it?
Every request row stores rate_card_id. You can join requests to rate_cards on that field to see the exact provider, model, and cost-per-token that produced any cost figure. It is visible in the request detail view in the dashboard.
Does Spineforge train on my prompts or data?
No. Spineforge acts as a secure transit and observability layer and does not use your prompts, responses or telemetry to train any AI models.
Is Spineforge SOC 2 / GDPR compliant?
Our infrastructure is being actively aligned with SOC 2, GDPR and the EU AI Act, and these efforts are in progress. We publish the current status of every item on the Security page rather than overstating it.
Can I self-host the gateway?
A self-hosted / on-prem gateway and dedicated single-tenant deployments are on the roadmap for Enterprise customers. They are not available yet. Talk to sales to shape what we build.
Is Spineforge open source?
Not currently. The architecture, metric calculations, and pricing formula are published openly because you should be able to understand exactly what you're running and paying for.
Does Spineforge support multi-tenant isolation?
Gateway keys and their data are scoped to your organization. Agents in different organizations cannot use each other's keys or access each other's telemetry. A dedicated single-tenant deployment is on the roadmap.
Is there a free plan?
Yes. The Free tier supports up to 3 agents and 10,000 gateway requests per month with no payment method required, including the key vault, budget caps, rate limits and the observability dashboard. See the Pricing page for details.
Don't see your question? Contact us: we'll answer and add it here.