Skip to content

SkillMeat v0.3.0-beta Release Notes

Release Date: November 17, 2025 Status: Beta Release Type: Feature Release

🎉 What's New

SkillMeat v0.3.0-beta is a major feature release that transforms SkillMeat from a CLI-only tool into a comprehensive platform for managing Claude artifacts. This release introduces a web interface, team sharing capabilities, MCP server management, marketplace integration, and enterprise-grade observability.

🌟 Highlights

Web Interface

  • Beautiful, responsive web UI built with Next.js 15 and App Router
  • Browse collections, deploy artifacts, and view analytics from your browser
  • Real-time sync updates via Server-Sent Events
  • Fully accessible (WCAG 2.1 AA compliant)
  • Dark/light mode support

Team Sharing

  • Export collections as signed .skillmeat-pack bundles
  • Share via Git, S3, or direct file transfer
  • Import with intelligent conflict resolution (merge, fork, skip)
  • Cryptographic Ed25519 signing and verification
  • Safe, idempotent imports

MCP Server Management

  • Deploy and configure MCP servers from CLI or web UI
  • Health monitoring with automatic status detection
  • Environment variable management with security warnings
  • Cross-platform support (macOS, Windows, Linux)
  • Easy deployment orchestration

Marketplace

  • Browse public marketplaces from multiple brokers
  • Install bundles with signature verification
  • Publish your own bundles with automated validation
  • License compatibility checking
  • Security scanning (40+ secret patterns, malicious code detection)

Enterprise-Ready

  • Comprehensive test coverage (85%+ backend, 75%+ frontend)
  • Structured logging with distributed tracing
  • Prometheus metrics with Grafana dashboards
  • Security review complete (A- rating, GA approved)
  • Load testing with 5 concurrent user scenarios

✨ New Features in Detail

1. Notification System

Real-time notifications for all operations with persistent history: - Bell icon notification center with unread badge - Multiple notification types: import results, sync results, errors, and info messages - Persistent history with localStorage (max 50 notifications) - Expandable details for import results showing success/failure breakdown - FIFO eviction automatically removes oldest notifications when limit reached - Accessibility fully keyboard navigable with screen reader support

2. Collections Navigation & Management

Browse and organize your artifact collections: - Collection switcher with persistent selection via localStorage - Collection context provider for easy access throughout the app - Full groups support - organize artifacts within collections into logical groups - Drag-and-drop reordering of artifacts within groups - ManageGroupsDialog for creating, editing, and deleting groups - CRUD API endpoints for programmatic group management

3. Groups Functionality

Powerful artifact organization system with full REST API: - Backend: SQLAlchemy models with Group and GroupArtifact relationship tables - API endpoints: - POST /api/v1/groups - Create new group - GET /api/v1/groups - List all groups with pagination - GET /api/v1/groups/{id} - Get group details - PUT /api/v1/groups/{id} - Update group metadata - POST /api/v1/groups/{id}/reorder - Reorder artifacts in group - DELETE /api/v1/groups/{id} - Delete group - Frontend: TanStack Query hooks with optimized cache invalidation - Features: Metadata support, artifact association, batch operations

4. Persistent Project Cache

Lightning-fast project loading with intelligent caching: - SQLite database at ~/.skillmeat/cache.db (replaces in-memory cache) - Thread-safe operations with proper connection pooling - Automatic background refresh with file system watcher - Models: Project, Artifact, Collection, GroupArtifact, Tag, Deployment - Performance: Project load time reduced from 1+ minutes to <100ms (95%+ cache hit rate) - Alembic migrations for schema evolution and versioning - Smart invalidation - automatic refresh on file changes detected by FSEvents/watchdog

5. Marketplace GitHub Ingestion

GitHub-backed marketplace with artifact ingestion workflow: - MarketplaceSource model for registering GitHub marketplace providers - Automated ingestion of artifacts from GitHub repositories - Search and filtering across marketplace sources - REST API endpoints for marketplace source CRUD operations - Web UI at /marketplace for browsing available artifacts

6. Agent Context Entities

Built-in artifact content hashing and validation: - Backend CRUD via Artifact model filtering with validation - Content-aware operations with SHA-256 hashing for integrity - CLI support via skillmeat context command group - Web UI at /context-entities for managing context artifacts - Validation ensures consistent artifact content before deployment

7. Artifact Flow Modal Redesign

Improved user experience for artifact operations: - Unified modal design for all artifact operations (create, edit, view) - Better error handling with inline validation feedback - Improved sync status tab showing real-time operation progress - Streamlined workflow for common operations

8. Tags Refactor v1

Complete tagging system with REST API and UI components: - CRUD API endpoints for tag management: - POST /api/v1/tags - Create tag - GET /api/v1/tags - List tags - GET /api/v1/tags/{id} - Get tag details - PUT /api/v1/tags/{id} - Update tag - DELETE /api/v1/tags/{id} - Delete tag - Artifact association - Associate multiple tags with artifacts - Frontend components: TagInput with autocomplete, tag filtering UI - Query hooks: useTagInput, useTagFilter with cache optimization - Bulk operations: Add/remove tags from multiple artifacts

📦 Installation

New Installation

pip install skillmeat==0.3.0-beta

Or with uv (recommended):

uv tool install skillmeat==0.3.0-beta

Upgrade from v0.2.0

See Migration Guide for detailed upgrade instructions.

All existing configurations and collections are fully backward compatible.

🚀 Quick Start

# Initialize collection
skillmeat init

# Start web interface
skillmeat web dev

# Add artifacts
skillmeat add anthropics/skills/code-review

# Deploy MCP server
skillmeat mcp add memory-server --repo anthropics/mcp-memory

# Browse marketplace
skillmeat marketplace-search python

📚 Documentation

🔒 Security

This release includes comprehensive security features:

Core Security - Ed25519 signature verification on all bundles (256-bit security) - SHA-256 hash verification before bundle import - OS keyring integration for credential storage - Rate limiting on all API endpoints (100 req/min per user)

Advanced Scanning - 40+ secret pattern detection (API keys, tokens, credentials) - Malicious code pattern scanning (eval, exec, dangerous imports) - Path traversal prevention in bundle extraction - CORS configuration hardened for security - Environment variable sanitization in MCP configs

Testing & Review - Comprehensive threat modeling with documented threat matrix - Security test suite with 98% coverage of critical paths - Penetration testing guide for pre-release assessments - Independent security review report (approved for GA)

🐛 Known Issues

Partially Implemented Features

  • Smart Import - Auto-populate from GitHub URLs functionality not yet complete (metadata extraction implemented, but URL-to-artifact mapping pending)
  • Versioning & Merge System - Core backend and API implemented, comprehensive testing deferred to v0.4.0

Other Known Issues

  • Zip bomb detection not yet implemented (scheduled for v0.4.0)
  • Dependency lock file generation pending (scheduled for v0.4.0)
  • MCP environment variable warnings could be more prominent
  • Single-region deployment (multi-region planned for v1.0)

Recent Fixes

  • LocalStorage SSR Compatibility - Fixed hydration warnings when using notifications and collection context in Next.js SSR
  • Collections API Endpoint Correction - Changed from /collections (read-only) to /user-collections for mutations (POST/PUT/DELETE)
  • Discovery Cache Invalidation - Fixed stale cache issues after artifact operations
  • Import Status Mismatch - Corrected distinction between failed imports (errors) and skipped imports (already exists)
  • Marketplace Re-import Fix - Deleting an imported artifact now resets the catalog entry status, allowing re-import. Added "Force Re-import" action for updating imported artifacts from upstream

See GitHub Issues for full tracking.

⚠️ Breaking Changes

API Changes

  1. Collections Endpoints
  2. Before: POST /collections for creating collections
  3. After: Use POST /user-collections for collection mutations (create/update/delete)
  4. Read-only: GET /collections remains read-only for browsing file-based collections
  5. Migration: Update any API clients or automation to use the new endpoint

  6. Groups API (New)

  7. New endpoints at /api/v1/groups for group management
  8. No migration needed as this is new functionality

  9. Cache Database Schema

  10. New SQLite database at ~/.skillmeat/cache.db
  11. Alembic migrations handle schema evolution
  12. Initial migration runs automatically on first startup

  13. CLI Context Command

  14. New command group skillmeat context for managing context entities
  15. No breaking changes to existing commands

No Deprecations

All existing features from v0.2.0 remain fully supported with backward compatibility.

🙏 Acknowledgments

Thanks to all beta participants for their invaluable feedback and contributions!

⚡ Performance Improvements

This release includes significant performance optimizations:

  • Project Load Time: Reduced from 1+ minute to <100ms with persistent cache
  • Cache Hit Rate: Achieving 95%+ hit rate for frequently accessed projects
  • Background Refresh: Intelligent file system watching reduces stale data
  • API Response Time: Sub-100ms response times for most endpoints
  • Memory Usage: Persistent cache reduces memory footprint

📅 What's Next (v0.4.0)

Feature Completions

  • Smart Import: Complete GitHub URL auto-population with metadata extraction
  • Versioning & Merge System: Full testing suite and comprehensive documentation
  • Zip bomb detection and validation
  • Dependency lock file generation for reproducible deployments

New Features

  • Cross-project search with advanced indexing
  • Smart updates with dependency tracking
  • ML-based recommendations for artifact discovery
  • Artifact provenance tracking

Focus Areas

  • Enhanced marketplace search and discovery
  • Improved team collaboration features
  • Advanced analytics and usage tracking

🆘 Support

  • Documentation: https://docs.skillmeat.dev
  • GitHub Discussions: https://github.com/miethe/skillmeat/discussions
  • Email: support@skillmeat.dev
  • Office Hours: Thursdays 2-3pm PT (during beta)

📝 Changelog

For complete changelog of all changes, see CHANGELOG.md.

🔄 Upgrade Path

From v0.2.0-alpha

  1. Update package:

    pip install --upgrade skillmeat==0.3.0-beta
    

  2. Verify installation:

    skillmeat --version
    

  3. Test web interface:

    skillmeat web dev
    

All existing collections, configurations, and deployments continue to work without changes.

From v0.1.0-alpha

Follow the v0.2.0 upgrade path first, then upgrade to v0.3.0 as above.

From skillman

Use the migration command:

skillmeat migrate --from-skillman

See migration documentation for details.

Database Migrations

This release introduces a new SQLite cache database. Migrations are handled automatically:

  1. First Startup: Alembic automatically runs initial migrations
  2. Subsequent Updates: New migrations apply automatically when needed
  3. Manual Check: View migration status with skillmeat cache status
  4. No Action Required: Most users don't need to manually manage migrations

Database Location: ~/.skillmeat/cache.db

What's Cached: - Projects and their metadata - Artifacts with deployment status - Collections and groups - Tags and classifications - Deployment configurations

For detailed migration help, see Cache Database Migration Guide.


Release Information: - Version: 0.3.0-beta - Release Date: November 17, 2025 - Type: Beta (stable API, new features, subject to feedback) - Python Support: 3.9+ - Node Support: 18+

Status: Ready for testing and feedback. Not recommended for production use without additional testing.