learning-recommendations

name: learning-recommendations description: "Use when presenting learning recommendations from background analysis, reviewing pending pattern suggestions, or approving automation scripts from traces. Load when learning-agent has findings to present."

Learning Recommendations

Review and act on recommendations from the background learning agent.

Use When

Workflow

Commands

List Pending

# List all pending recommendations
python3 ~/.claude/skills/learning-loop/scripts/learning-agent.py --list-pending

# Check if any exist
ls ~/.claude/learning-recommendations/recommendations_*.json 2>/dev/null

Process Recommendation

For each recommendation, use AskUserQuestion:

AskUserQuestion(
    questions=[{
        "question": "New Pattern Detected: {pattern}\n\nFound {count} occurrences. Add to skill references?",
        "header": "Learning",
        "options": [
            {"label": "Add to skill refs", "description": "Update skill with new pattern"},
            {"label": "Skip", "description": "Don't add now"},
            {"label": "Reject", "description": "Not a valid pattern"}
        ],
        "multiSelect": false
    }]
)

Clear Processed

# Clear all processed recommendations
python3 ~/.claude/skills/learning-loop/scripts/learning-agent.py --clear

Recommendation Types

Storage

Directory: ~/.claude/learning-recommendations/

Format: recommendations_YYYYMMDD_HHMMSS.json

{
  "timestamp": "2026-01-03T22:42:27",
  "recommendations": [
    {
      "type": "pattern",
      "title": "Pattern: deployment",
      "description": "Found 3 occurrences...",
      "data": {...},
      "actions": [...]
    }
  ]
}