Skip to content

Developer Portal Integration

This guide describes how developers discover, browse, and deploy SkillMeat artifacts directly within Backstage's developer portal surface.

Overview

SkillMeat's Backstage plugin surfaces artifact discovery and deployment capabilities in two key developer workflows:

  1. Entity Pages — View artifact deployment status, versions, and health directly on your service catalog entries
  2. Scaffolder Templates — Provision new projects with pre-loaded AI context and automation workflows

This guide covers the developer experience of using these surfaces. For platform engineer setup, see Catalog Integration Guide.

Prerequisites

Before using SkillMeat in Backstage, ensure:

  • The SkillMeat Backstage plugin is installed and registered in your Backstage app
  • Your organization's artifact catalog is connected via the SkillMeat catalog provider
  • You have access to the Backstage catalog and scaffolder
  • Your organization has deployed SkillMeat artifacts to share (skills, commands, agents, context packs, etc.)

Accessing the Artifact Portal

Entity Page Integration

SkillMeat appears as a tab on your service's entity page within Backstage's catalog.

Navigation:

  1. Go to Catalog in your Backstage sidebar
  2. Search for and open your service or project entity
  3. Click the SkillMeat tab (if present; requires platform engineer setup)

You'll see three cards summarizing your deployment's artifact state:

Card Shows
Context Card Deployment context, bundle version, and sync status
BOM Card Bill of materials — list of deployed artifacts with signature status
Drift Card File-level drift detection highlighting changes since last sync

Intelligence Tab (Optional)

If your organization enables telemetry analytics, an Intelligence tab appears combining context cards with workflow effectiveness metrics and agentic performance signals.

Browsing Artifacts

Overview

The SkillMeat plugin integrates with Backstage's catalog discovery. Artifacts appear as discoverable entities within your catalog, alongside services, libraries, and other IDP assets.

Artifact Types You May Encounter:

  • Skills — Reusable Claude code capabilities (canvas design, analysis, etc.)
  • Commands — CLI automation (deployment, validation, etc.)
  • Agents — Specialized subagents with pre-configured knowledge
  • Workflows — Multi-step automation orchestrations
  • Context Packs — Pre-built AI context for specific domains (compliance, fintech, etc.)
  • Bundles — Curated collections of related artifacts

Discovering Artifacts in the Catalog

Artifact Visibility

Artifacts must be registered with the SkillMeat catalog provider by your platform team. See Catalog Integration Guide for provider setup.

  1. Navigate to Catalog → search bar at top
  2. Search by:
  3. Artifact name (e.g., "canvas design", "python refactor")
  4. Type filter — narrow to Skills, Commands, Agents, etc.
  5. Owner or team — find artifacts maintained by specific teams
  6. Tags — filter by domain (e.g., compliance, fintech, performance)

  7. Click an artifact to view its detail page

  8. You'll see:
  9. Description — What the artifact does
  10. Versions — Available versions and release history
  11. README — Full documentation and usage examples
  12. Dependencies — Other artifacts this one requires
  13. Ownership — Team or person responsible

Viewing Artifact Details

Each artifact detail page contains:

Header Section: - Artifact name, type badge, and owner - Star/favorite button (to save for quick access) - Version selector dropdown

Tabs:

Tab Contains
Overview Artifact description, tags, and key metadata
Readme Full documentation and usage instructions
Versions Release history with changelog entries and available versions
Dependencies Required and recommended artifacts
Deployments Which projects currently use this artifact

Actions: - Deploy button — Deploy this artifact to your project (see Deploying Artifacts below) - View Source — Link to GitHub repository or source - Copy Identifier — Quick copy of artifact reference for use in scripts

Deploying Artifacts

Self-Service Deployment

Developers can deploy artifacts directly from the catalog to projects they own.

Step 1: Locate the Artifact

  1. Navigate to Catalog and search for the artifact you need
  2. Click the artifact to open its detail page

Step 2: Initiate Deployment

Click the blue Deploy button on the artifact detail page.

A deployment modal opens with options:

Option Description
Target Project Select which project to deploy to (dropdown; shows your accessible projects)
Deployment Method Local (write directly to .claude/) or Pull Request (create a PR for team review)
Branch Name (PR method only) Name of the feature branch (e.g., chore/add-canvas-skill)
PR Title (PR method only) Title for the pull request on GitHub
Description (PR method only) Optional markdown body for the PR
Reviewers (PR method only) Comma-separated GitHub usernames to request review

Step 3: Confirm Deployment

Deployment Impact

Deploying to a project writes artifact files to .claude/ in that project. Local deployments take effect immediately; PR deployments require code review and merge.

Review your selections and click Deploy or Create Pull Request.

Local Deploy Output:

✓ Deployed canvas-design to my-service-project
  Files: skill.md, README.md, templates/ 
  Location: ./.claude/skills/canvas-design/

PR Deploy Output:

✓ Pull request created
  URL: https://github.com/my-org/my-service-project/pull/42
  Branch: chore/add-canvas-skill
  Files: 1 artifact, 3 files changed

Using Scaffolder Templates

Scaffolder templates provide a guided project provisioning workflow that can inject SkillMeat artifacts automatically.

Triggering a Template

  1. Navigate to Create in the Backstage sidebar
  2. Browse or search for scaffolder templates
  3. Look for templates tagged with skillmeat or ai-context
  4. Click a template to begin the scaffolding wizard

Template Steps

Templates typically include:

Step 1: Project Details - Project name and description - Owner (team or individual) - Repository URL (for Git-based projects)

Step 2: AI Context Selection (SkillMeat-enabled templates) - Available context packs and skill bundles - Checkboxes to select which artifacts to inject - Organization policies and required artifacts (pre-selected)

Step 3: Advanced Options - Git branch naming - Custom configuration - Integration options (CI/CD, compliance, monitoring)

Artifact Injection During Scaffolding

Once you complete the template wizard, the scaffolder:

  1. Creates your project repository on GitHub
  2. Runs the SkillMeat scaffolder backend action to inject selected artifacts
  3. Writes artifacts to .claude/ in the new repository
  4. Configures deployment settings if applicable
  5. Optionally creates initial PR with context loaded

Example Output:

✓ Project created: my-fintech-service
  Repository: https://github.com/my-org/my-fintech-service

✓ AI Context injected:
  - compliance-context-pack/
  - python-refactor-skill/
  - regulatory-validator-command/

✓ Initial PR created with artifacts
  URL: https://github.com/my-org/my-fintech-service/pull/1

Integration with CI/CD

Artifact Status in Deployments Tab

If your SkillMeat integration tracks deployments, you can monitor:

  • Deployment history — Which artifacts were deployed when
  • PR status — Open, merged, or closed deployment PRs
  • Sync status — Whether deployed artifacts are up-to-date with source
  • Drift detection — Files that have changed locally vs. the deployed version

This information appears in a Deployments tab on your entity page (requires telemetry integration).

Using Artifacts in Workflows

Deployed artifacts are immediately available to Claude Code within the project:

# After deploying, artifacts are in .claude/
ls -la ./.claude/skills/
ls -la ./.claude/commands/

# Reference in Claude sessions
# Skills and commands activate automatically in Claude Code

Artifacts become active context for any Claude AI session operating on the project.

Search and Discovery

Use Backstage's search to find artifacts:

Search Syntax:

# By artifact type
type:skill
type:command
type:agent
type:workflow
type:context-pack

# By tag
tag:compliance
tag:fintech
tag:performance

# By owner
owner:platform-team
owner:alice

# Combined
type:skill tag:analysis owner:platform-team

Filtering in Artifact Views

In the catalog grid view, use filter controls to:

  • Type — Show only specific artifact types
  • Tag — Filter by domain or category
  • Owner — Find artifacts by team ownership
  • Status — Show only available (vs. deprecated or archived) artifacts
  • Scope — Show user scope (global) or local scope (project-specific) artifacts

Viewing Deployment Context

Context Card

The Context Card on your entity's SkillMeat tab shows:

  • Project ID — Unique SkillMeat deployment identifier
  • Deployment Bundle — Active bundle version
  • Sync Status — Whether deployed artifacts match upstream sources
  • Last Sync — Timestamp of last sync operation
  • Artifact Count — Number of deployed artifacts

Bill of Materials (BOM) Card

The BOM Card lists:

  • Artifacts deployed — Name, type, version for each artifact
  • Signature status — Whether each artifact is signed (security compliance)
  • Artifact versions — Which version of each artifact is deployed
  • Metadata — Tags, owner, scope for each artifact

Example BOM view:

Deployed Artifacts: 5

✓ canvas-design (v2.1.0)          Signed
✓ python-refactor (v1.3.0)         Unsigned
✓ compliance-context-pack (v3.0.0) Signed
✓ my-custom-skill (local)          Unsigned
✓ validation-workflow (v1.0.0)      Partial

Drift Card

The Drift Card shows:

  • Drift status — Whether any deployed files have changed locally
  • Changed files — Line-by-line breakdown of modifications
  • Drift severity — Critical (modified logic), Minor (docs/comments), None

Example:

Drift Detected: 3 files

✗ .claude/skills/canvas-design/SKILL.md (Minor)
  Lines 5-8 modified

✗ .claude/commands/deploy/cli.py (Critical)
  Function signature changed on line 42

⚠ .claude/context-pack/compliance/README.md (Minor)
  Comments updated

Accessibility

Keyboard Navigation

All SkillMeat portal surfaces support full keyboard navigation:

  • Tab — Move between controls
  • Enter/Space — Activate buttons, toggle checkboxes, select options
  • Arrow keys — Navigate dropdown options, table rows
  • Escape — Close modals and dropdowns

Screen Reader Support

SkillMeat cards and deployment modals include:

  • Semantic HTML with <button>, <table>, <form> landmarks
  • ARIA labels for icon-only buttons and status indicators
  • Table headers with proper association to data cells
  • Form field labels and validation error announcements

Screen Reader Testing

SkillMeat artifact portal supports major screen readers (NVDA, JAWS, VoiceOver). Report accessibility issues to your platform team.

Troubleshooting

SkillMeat Tab Not Visible

Problem: The SkillMeat tab doesn't appear on your entity page.

Solutions: - Confirm your entity has the skillmeat.io/deployment-set-id or skillmeat.io/project-id annotation (added by catalog provider) - Verify the SkillMeat plugin is registered in your Backstage app-config.yaml - Check your browser console for JavaScript errors (F12 → Console tab) - Refresh the page (Ctrl+Shift+R)

Artifacts Not Showing in Catalog

Problem: You search the catalog but SkillMeat artifacts don't appear.

Solutions: - Verify the SkillMeat catalog provider is configured (see Catalog Integration Guide) - Check that artifacts are actually registered with SkillMeat - Try searching by artifact name directly instead of filters - Ask your platform team to verify the provider sync completed: skillmeat catalog refresh

Deploy Modal Fails to Open

Problem: Clicking Deploy shows an error or the modal won't open.

Solutions: - Check that the SkillMeat API is reachable from your Backstage instance - Verify the skillmeat.baseUrl in app-config.yaml is correct - Check browser console (F12 → Console) for network errors - Confirm you have permission to write to the target project - If using PR deploy, verify your Git credentials are configured

Deployment Hangs or Times Out

Problem: Clicking Deploy seems to hang, or you see a timeout error.

Solutions: - Network connectivity — verify Backstage can reach SkillMeat API and GitHub - Large artifacts — artifacts over 10MB may take longer; wait a bit longer - Token issues — for PR deploy, confirm your GitHub PAT is valid and has repo scope - Check platform team logs for backend errors: docker logs skillmeat-api (if containerized)

Permission Denied on PR Deploy

Problem: Deploy via PR fails with "Permission denied" or "Unauthorized".

Solutions: - Missing token: Ensure you've configured GitHub authentication in Backstage - Insufficient scopes: Your GitHub PAT needs the repo scope (not just public_repo) - Repository access: Confirm you have write access to the target repository - Archived repository: Some orgs archive inactive repos — verify the repo is not archived

Best Practices

Choosing Deployment Method

Method When to Use When NOT to Use
Local Deploy Testing new artifacts locally; small experimental changes Production environments; team-shared projects; regulated domains
PR Deploy Team environments; requiring code review; compliance-regulated projects Rapid prototyping; learning/exploration

Artifact Selection

  • Start small — Deploy one artifact at a time to understand its impact
  • Read documentation — Check the artifact's README before deploying
  • Review dependencies — Some artifacts require others (check Dependencies tab)
  • Check versions — Prefer stable versions (v1.0.0) over beta tags (v1.0.0-beta)

Managing Deployed Artifacts

  • Monitor drift — Check the Drift card regularly; merged changes may diverge from upstream
  • Update regularly — New versions include bug fixes and improvements
  • Remove unused artifacts — Undeploy artifacts your project no longer uses
  • Document customizations — If you modify an artifact locally, record why (in drift comments)

Next Steps

Support & Feedback

For issues or questions about the SkillMeat developer portal:

  • Technical issues: File an issue in your organization's SkillMeat tracking
  • Feature requests: Contact your platform engineering team
  • Documentation feedback: Submit corrections via your feedback channel