v0.1.2

Added

• New knowledge skill: tool-uv-monorepo - Comprehensive guide for building Python monorepos with uv workspaces
  • Unified dependency resolution with single uv.lock
  • PEP 735 dependency groups for dev tools
  • Pytest importlib mode for test isolation
  • Docker multi-stage builds with uv export
  • GitHub Actions global cache strategy
  • Migration guide from poly-repo to monorepo

v0.1.1

Fixed

• Resolved plugin validation issues identified in comprehensive review
• Added missing ml-experiment example and template files
• Fixed agent color conflict: config-generator changed from magenta to green
• Removed empty ml-model-export/scripts directory

Changed

• Implemented progressive disclosure pattern for large knowledge skills
• Reduced tool-pixi skill from 1560 lines to 600 lines (62% reduction)
• Reduced ml-transformers skill from 1182 lines to 550 lines (53% reduction)
• Reduced ml-cli-tools skill from 1093 lines to 350 lines (68% reduction)
• Moved advanced topics to reference/ subdirectories for better organization
• Total documentation reduction: 2,335 lines (62% reduction)

Improved

• Plugin structure now follows progressive disclosure best practices
• Context window usage optimized for Claude Code
• Skill organization improved with reference files for detailed content
• Overall plugin quality score improved from 8.5/10 to 9.5/10

v0.1.0

Added

• Initial release of ML Research plugin
• 12 workflow skills for ML development:
  • /ml-train - Execute training runs
  • /ml-config-manager - Generate Hydra configs
  • /ml-debug - Debug training issues
  • /ml-experiment - Manage experiments
  • /ml-validate - Validate project structure
  • /ml-profile - Profile performance
  • /ml-data-pipeline - Manage data pipelines
  • /ml-setup - Setup environment
  • /ml-project-init - Initialize projects
  • /ml-lint - Code quality checks
  • /ml-format - Format code
  • /ml-model-export - Export models
• 8 knowledge skills:
  • ml-lightning-basics - PyTorch Lightning patterns
  • ml-hydra-config - Hydra configuration
  • ml-pytorch-geometric - Graph Neural Networks
  • ml-wandb-tracking - Experiment tracking
  • ml-transformers - Hugging Face Transformers
  • ml-cli-tools - Building CLIs with Typer/Rich
  • tool-pixi - Pixi package manager
  • tool-marimo - Marimo reactive notebooks
• 6 specialized agents:
  • ml-architect - Design ML architectures
  • training-debugger - Diagnose training issues
  • config-generator - Generate Hydra configs
  • pytorch-expert - PyTorch optimization
  • geometric-specialist - Graph Neural Networks
  • transformers-specialist - LLM fine-tuning
• Comprehensive documentation in CLAUDE.md
• GitHub Actions CI/CD workflows
• Pre-commit hooks for code quality
• Plugin marketplace integration

Documentation

• Detailed CLAUDE.md with architecture overview
• Git workflow and commit conventions
• Frontmatter reference for skills and agents
• Common patterns and pitfalls guide