/advise — Skills Registry Lookup

Search the Skills Registry for relevant experiments before starting new work.

Steps

1. •Read the user's goal from the conversation context
2. •Search Skills_Registry/plugins/ for related skills by:
   • •Scanning description fields in each plugin.json (look for matching trigger conditions)
   • •Reading SKILL.md files in matching plugins for detailed findings
   • •Check all categories: general/, scientific/, trading/, kintsugi/, maxfuse/, training/
3. •For each relevant skill found, summarize:
   • •What was tried and what worked (from "Verified Workflow" section)
   • •What failed and why (from "Failed Attempts" table — the most valuable section)
   • •Recommended parameters (from "Final Parameters" section)
   • •Environment/version notes
4. •If no relevant skills found:
   • •Inform the user that no prior experiments match their goal
   • •Suggest running /retrospective after completing their task to capture learnings

Where to Search

The Skills Registry is a git submodule at Skills_Registry/:

Skills_Registry/plugins/
  ├── general/              # Cross-project Python/dev skills
  ├── scientific/           # Scientific computing patterns
  ├── trading/              # Alpaca Trading system skills
  ├── kintsugi/             # KINTSUGI-specific skills
  ├── maxfuse/              # MaxFuse imaging integration
  ├── training/             # Training experiment documentation
  └── templates/            # Skill templates

Each plugin has:

• •.claude-plugin/plugin.json — metadata with trigger conditions in description
• •skills/<name>/SKILL.md — detailed documentation with verified workflows and failed attempts

You can also consult Skills_Registry/marketplace.json for a quick index of all plugins.