Compact

Consolidate clusters of small related notes into canonical files while preserving content and wikilinks.

Parameters

• target path (default: current working directory)
• --mode report|apply-safe|apply (default: report)
• --yes execute without confirmation
• --max-lines N threshold for small files (default: 50)
• --min-cluster N minimum files per cluster (default: 2)

Scope

Scan markdown recursively in target path, excluding hidden/system directories and all raw/* paths (raw/sources/, raw/processed/, raw/assets/).

Clustering

Prioritize these signals:

1. Date-series basename matches (strongest)
2. High-overlap tags + similar topic in same area
3. Small sibling fragments in same folder

Confidence Rubric

• high: strong date-series/basename match and low link rewrite risk
• medium: tag/folder similarity with manageable rewrite risk
• low: weak semantic similarity or ambiguous destination

Workflow

1. Discover small-note candidates under threshold.
2. Build clusters and assign confidence.
3. Plan destination files and wikilink rewrites.
4. Generate preview plan.
5. Execute by mode:
   • report: plan only
   • apply-safe: execute high confidence clusters only
   • apply: execute all user-confirmed clusters
6. In write modes, confirm unless --yes.
7. Execute merge:
   • merge content with one canonical frontmatter block
   • preserve content (strip only duplicate per-file frontmatter/H1)
   • update wikilinks
   • archive originals to mirrored archive/ paths
8. Append operation entry to log.md.

Safety

• Default to report-first.
• Do not hard-delete originals unless explicitly requested.
• Verify merged output contains all source entries before archiving originals.
• Keep operations reversible and report touched files.

Output

Return:

• clusters by confidence
• selected execution set by mode
• destination files
• link updates applied/proposed
• archive moves applied/proposed
• touched files