Structure Text into Hierarchical Bullet Points

Bulletmind converts any input - dense prose, articles, messy notes, or inconsistent existing bullet lists - into a clean, strictly hierarchical bullet structure with no paragraphs and no prose bridging lines. It operates at three intensity levels, switchable with /bulletmind lite|full|ultra: lite does light restructuring while preserving sentence flow, full (the default) applies strict hierarchy with balanced compression and clear grouping/splitting, and ultra applies deep hierarchical decomposition with aggressive splitting for maximum structural granularity. The resulting structure follows consistent indentation:

- Top-level idea
  - Sub-point
    - Detail
  - Sub-point
- Next top-level idea
  - Sub-point

Use it when the user wants a bullet-only summary of dense text, notes, explanations, articles, or webpages; cleaned-up note-taking with clear parent-child relationships; structured study material that's easier to scan and memorize; or consistent formatting applied to a messy or mixed bullet list. Do not use it when the user requests paragraphs, for creative writing tasks like stories or essays, or in any format where bullets would reduce clarity or violate a requested output format (tables, code blocks, JSON, prose) - a higher-priority instruction for one of those formats overrides the skill's bullet-only behavior.

Input can be a paragraph, existing bullets, or even short text - all get converted into a bullet tree. The transformation logic is consistent regardless of source: a paragraph's main ideas become top-level bullets with details nested beneath; messy notes get cleaned into a proper hierarchy; existing bullets get restructured and their depth normalized rather than left as-is; and even short input still gets converted into a (small) bullet tree rather than left as plain text. Formatting rules: only - bullets, 2-space indentation per level, one idea per line, no mixed symbols. Compression rules: remove filler words, split complex sentences, preserve key facts and relationships, but never flatten the hierarchy down to one level and never over-summarize past what the source supports - clarity is prioritized over maximum compression. When active, the output is the bullet hierarchy itself, with no added commentary or explanation.

Works across Claude, Cursor, Gemini, and Codex per its tool tags; output templates are documented in a companion EXAMPLES.md file.

Anyone taking notes, studying dense material, or cleaning up messy bullet lists who wants a consistent, scannable hierarchy instead of manually reformatting text by hand. The skill explicitly prioritizes clarity and a logical tree structure over maximum compression, and avoids flattening everything into one level.