StackSpine

Control Production AI

6 followers

Control Production AI

6 followers

Open-source (Apache 2.0) control plane and system of record for production AI: policy enforcement, budget guardrails, compliance, and audit across every provider.

Free Options

Launch tags:Productivity•Open Source•Artificial Intelligence

Launch Team

SkydiveAI coworkers that actually get work done

Promoted

Maker

📌

StackSpine – an open-source proxy for LLM routing and budget checks StackSpine is an open-source proxy for LLM apps. It sits between an app and model providers so requests can be routed, budget-checked, and logged before they are sent upstream. Repo: https://github.com/stackspine/ga...  Demo: https://stackspine.com/demo/flow License: Apache-2.0 I built StackSpine because directly calling providers are easy in the beginning. I kept running into questions like: * which app or task is spending money? * should this request be blocked before it hits a provider? * what happens if a provider is unhealthy? * where do routing decisions get recorded? * how do I avoid rewriting code every time I add or change providers? What works in the open-source repo today: * gateway/proxy for LLM requests * provider routing * budget checks before dispatch * provider health checks and fallback routing * structured request logs * local configuration for routes, providers, budgets, and policies * SDK/API examples * Docker/local development setup The hosted demo is just a visual walkthrough. The thing to try is the repo.

Report

2mo ago

A first-class Python SDK would be a huge win since most of the AI ecosystem is Python-first. Even a thin wrapper that mirrors the existing REST API and handles auth + retries would make StackSpine feel native to the tooling folks already have in production.

Report

1mo ago

Maker

@mirayj8ur Thanks for your comment!

Report

1mo ago

Maker

I built StackSpine because production AI apps quickly become hard to control once teams use multiple providers, models, apps, and workflows.

StackSpine is an open-source control plane that helps teams route AI requests, enforce budget and policy rules before execution, and keep a structured record of what happened.

The open-source repo includes the gateway/proxy, provider adapters, budget and policy checks, provider health/fallback routing, structured logging, SDK/API examples, and a Docker-based local setup. No signup is required to run it locally.

I’d love feedback from anyone building or operating AI systems in production, especially around routing, cost control, governance, and observability.

Report

2mo ago

Reviews