AI Explorer - Spotify for Backstage

AI Explorer is a centralized marketplace for agent resources that make AI coding agents more effective across your organization. Instead of every team writing their own agent instructions from scratch, teams publish their expertise as reusable resources that others can discover and install.

Why AI Explorer?

AI coding agents like Claude Code, Cursor, and Codex work best when they have context about your organization’s standards, workflows, and tools. AI Explorer solves the distribution problem:

Discover what’s available by browsing a searchable marketplace of rules and skills
Install resources through the Portal UI or your agent’s plugin system
Contribute your team’s expertise so others can benefit from it
Certify high-quality resources so developers know what to trust

Key concepts

Concept	Description
Rule	A short guideline that agents load into context — e.g. “always use design tokens for UI components”
Skill	A multi-file instruction set that teaches agents a specific workflow — e.g. “how to create a migration”
Plugin	A package that wraps one or more skills so they can be shared and installed together
Marketplace	A catalog of plugins hosted in a Git repository. Users add a marketplace to browse and install plugins

See the individual pages for details on each:

Rules: guidelines and standards for AI agents
Skills: workflow instructions for AI agents
Marketplaces & Plugins: grouping and distributing resources

How it works

Resources live as files in your GitHub repositories. AI Explorer’s entity providers scan configured GitHub organizations and automatically discover rule files and SKILL.md files using GitHub code search. The providers parse each file’s frontmatter to build AiContext entities in the Backstage catalog — no catalog-info.yaml is required.

GitHub Repos (source of truth)
       ↓
Entity Providers (scan for rule files & SKILL.md)
       ↓
Backstage Catalog (AiContext entities)
       ↓
Portal UI (browse & search)

Browsing resources

AI Explorer provides dedicated pages for browsing and searching resources. You can filter by:

Type — rules, skills, or plugins
Discipline — e.g. backend, web, android, data (rules only)
Category — e.g. testing, code-review, architecture
Owner — the team that maintains the resource
Certification status — whether the resource has been reviewed and certified
Lifecycle — experimental, production, deprecated

Each resource has a detail page showing its content, metadata, and install instructions.

What’s Next?

In the following sections, instructions will be provided that will guide you on how to:

Configure AI Explorer in the Config Manager
Write effective Skills and Rules
Install and use AI Explorer resources
Configure soundcheck for AI Explorer to ensure Skill Quality

​Why AI Explorer?

​Key concepts

​How it works

​Browsing resources

​What’s Next?

Why AI Explorer?

Key concepts

How it works

Browsing resources

What’s Next?