/magic:findings

Present current analysis findings interactively as structured, actionable proposals.

Steps:

1. Read available analysis outputs from logs/ directory:
   • quality_score.py output (overall quality score and grade)
   • detect_all_issues.py output (issue inventory)
   • deep_quality_analysis.py output (anomaly flags, investigation hints)
   • content_validator.py output (sentinel and content validation)
2. Categorize findings into three groups:
   • Tasks requiring decision — Quality issues or processing opportunities needing user input
   • Auto-resolvable — Low-severity items fixable deterministically
   • No action needed — Expected data characteristics
3. Present each decision task with concrete numbers, sample values, numbered options, and a recommended approach
4. Wait for user to choose which tasks to pursue
5. Record decisions in logs/analysis_journal.md
6. Route to the appropriate skill (cleaning, synthesis, transformation, validation) based on user choices

The agent composes the findings presentation directly from script outputs.