| name | celebrity-distiller |
| description | Distill a person's biography, book content, or name into a comprehensive decision-making profile and actionable playbook. Extracts core values, mental models, principles, and situational rules from any source โ biography, autobiography, or book excerpts. |
| license | MIT |
Celebrity Distiller
Given any source about a person โ their name, a biographical paragraph, or excerpts from a book โ produce a comprehensive decision-making profile and actionable playbook. Use the source to understand how they think, what they value, and how they decide.
Input Handling
Determine the input type and behave accordingly. Do NOT use web search or external tools โ only what the user provides or your training knowledge.
Name only (well-known person):
Produce a full draft profile and playbook from your training knowledge. Tag every item [general knowledge]. After outputting, prompt the user to paste book excerpts to strengthen the profile.
Name only (obscure or uncertain):
If you cannot produce at least 3 distinct, grounded principles with reasonable confidence, do not guess. Say so clearly and ask the user to paste source material before proceeding.
Bio or Wikipedia text:
Extract directly from the provided text. Ground every principle in what was stated.
Book excerpts (primary use case):
Treat excerpts as primary source material. Quote directly. Ground all principles in specific passages.
Mixed or multi-round input:
Merge new content into the existing profile. Do not restart from scratch.
Source Confidence Markers
Tag every principle, behavior, and playbook item with one of:
[from source] โ directly stated in the provided material[inferred] โ reasonably derived from patterns in the material[general knowledge] โ from training data, not the provided text[conflicting sources] โ new material contradicts an existing point; present both versions
Retain all markers in the final consolidated output.
Part 1 โ Comprehensive Profile
After determining input type and gathering enough material, produce the following structure. Target a profile readable in 10 minutes. Aim for depth over breadth โ 3-5 strong, well-evidenced principles are better than 10 thin ones.
Core Values
3-5 deeply held beliefs that drove their decisions. Each value:
- stated as a short label (e.g., "Perfectionism over shipping speed")
- followed by 1-2 sentences of explanation
- tagged with a source confidence marker
Mental Models
How they thought about problems. Examples: first principles reasoning, long-term vs short-term framing, systems thinking, etc. Each model:
- named and briefly explained
- grounded in a specific decision or behavior from the source
- tagged with a source confidence marker
Decision-Making Principles
Rules they applied consistently across contexts. Each principle:
- stated as a concrete rule (not an abstraction)
- illustrated with an example from their life or work
- tagged with a source confidence marker
Signature Behaviors
Patterns others observed โ how they ran meetings, hired people, handled failure, responded to criticism, etc. Each behavior:
- described specifically (not "he was detail-oriented" but "he reviewed every line of ad copy before it ran")
- tagged with a source confidence marker
Key Quotes
Direct quotes from the source material, each tagged with a source confidence marker, anchoring the principles above.
Format: "[Quote]" โ [context of when/where it was said] [source confidence marker]
If input was name-only, paraphrase known quotes and tag [general knowledge].
Part 2 โ Actionable Playbook
Immediately after Part 1, produce the playbook. Derive it from the profile โ do not introduce new principles here.
Playbook items must be concrete rules, not vague abstractions.
- Good: "Would cut the feature scope, not the timeline."
- Bad: "Prioritized quality."
Each item carries a source confidence marker.
The five default situational headers below are starting scaffolding. For subjects where these frames don't fit (e.g., a musician, scientist, or athlete), replace them with more relevant situations derived from the source material.
When facing a difficult decision...
[1-3 concrete rules, each tagged with a source confidence marker]
When dealing with failure...
[1-3 concrete rules, each tagged with a source confidence marker]
When evaluating people or talent...
[1-3 concrete rules, each tagged with a source confidence marker]
When setting priorities...
[1-3 concrete rules, each tagged with a source confidence marker]
When under pressure...
[1-3 concrete rules, each tagged with a source confidence marker]
[Add person-specific sections as warranted by the source material]
Refinement Loop
After outputting Part 1 and Part 2, prompt the user:
Want to refine this? You can:
1. Paste more excerpts from the book
2. Correct something that feels off
3. Add a specific domain (e.g., "focus on how they managed teams")
4. Type 'done' to finish
For each refinement round:
- New excerpts: Extract new principles or behaviors and merge them into the existing profile. Do not restart.
- Corrections: Update the flagged item and note what changed.
- Domain focus: Deepen the relevant subsection(s) with more specificity.
- Contradictions: If new material conflicts with an existing point, flag it
[conflicting sources] and present both versions side by side.
After each round, reprint the affected subsections with [updated] or [added] markers. Full reprinting is acceptable โ the markers are best-effort annotations, not guaranteed minimal diffs.
When the user types done:
Reprint the complete, consolidated profile โ both Part 1 and Part 2 in full โ with all source confidence markers retained. This is the final reference document.
