Backstage + AI: Stop Hunting for AI Assets

Julian Pedro F. Braga — Sun, 15 Mar 2026 16:35:36 +0000

If you're working in an engineering org that uses Backstage, you already know the "Golden Path" philosophy: everything you need — infra, docs, services — lives in one unified catalog. Backstage solved the microservices mess, but now we're facing a new kind of chaos: AI fragmentation.

The Problem: "Where Did Team B Save That Agent?"

Right now, nearly every squad is busy spinning up their own GitHub Copilot instructions and specialized agents. The catch? All that institutional knowledge ends up trapped in siloed .md files or buried in some obscure repo.

When you switch projects or need to follow an architecture standard another team has already figured out, you inevitably reinvent the wheel.

Squad Silos — Two teams writing the same instructions for the same stack.
Zero Visibility — No way to know an agent already exists to automate that tedious internal process.
Context Drifting — Keeping security and coding standards consistent through AI becomes impossible without a single source of truth.

The Solution: Dev AI Hub for Backstage

The Dev AI Hub is a plugin that treats your AI-native assets — Copilot instructions, specialized agents, and Claude configurations — as first-class citizens in your Backstage catalog.

The concept is simple: if it's an engineering asset, it belongs in the catalog. This way, any developer in your company can discover and reuse what's already been validated.

The Game Changer: Built-in MCP (Zero Context Switching)

We know devs hate jumping between tabs. That's why the most powerful feature of the Dev AI Hub is its Model Context Protocol (MCP) support.

The plugin acts as a centralized MCP Server. This means your AI tools — Claude Code, Gemini, and GitHub Copilot — can talk directly to your Backstage catalog without you ever leaving the IDE.

Imagine asking your terminal:

"What are the standard instructions for our Kubernetes deployments?"

"Use the security agent from the Cloud team to review this PR."

Your AI assistant fetches that live, up-to-date context straight from Backstage — in real time.

Why You'll Love It

IDE-Native Context — Use MCP to give your AI tools (Claude, Copilot, etc.) a "brain" filled with your company's actual, living data.
No More Copy-Paste — Stop copying instruction files from repo to repo. Centralize them once and reference them everywhere.
Real Collaboration — Discover agents built by other teams that can automate your current workflow.
Enforced Standards — Ensure everyone uses the same architectural and security guidelines through AI.

Try It Out

Stop hunting for config files and start scaling your team's collective intelligence.

👉 Check the plugin on GitHub: backstage-dev-ai-hub

⭐ If you find it useful, give it a star — it helps others discover the project!