philosophy-distiller

Philosophy Distiller

This skill turns a pile of real writing into a living PHILOSOPHY.md.

It is not just extraction. It is distillation + refinement:

read a large corpus
separate recurring convictions from context noise
draft a philosophy document
identify recurring bullshit patterns, evasions, and coherence failures
crystallize the worldview into memorable, evidence-earned lines
refine that document as more samples, corrections, and edge cases appear

Use it when the user says things like:

“Here’s a dump of my essays—figure out what I keep believing.”
“Extract my worldview from all this writing.”
“Refine this PHILOSOPHY.md based on more samples.”
“Tell me what I’m always arguing for and against.”
“Map the themes, values, tensions, and obsessions that actually recur.”

Core Philosophy

The job is not to psychoanalyze the writer. The job is to compress recurring reality.

That means:

prefer repeated commitments over flashy one-off declarations
separate worldview from topic, platform, mood, and debate-specific rhetoric
notice where the writer regularly cheats, hedges, overreaches, or hides
when helpful, connect the worldview to existing thinkers, traditions, or concept clusters instead of reinventing familiar ideas in flatter language
mark weak signals as tentative instead of pretending certainty
refine the document in-place instead of rewriting it from scratch every pass

The result should feel like:

“Yeah, that’s actually what I keep trying to say about the world.”

and also:

“Yeah, fuck, that is exactly how I bullshit when I haven’t fully earned the claim.”

Not:

“This is a Pinterest mood board for my opinions.”

Default Mode

Default behavior:

If a PHILOSOPHY.md already exists, refine it.
If no PHILOSOPHY.md exists, create a first draft from the corpus.

This skill is meant for a long, thoughtful session where extraction and refinement happen together.

Inputs

Collect or infer these before distilling:

Writing corpus — essays, articles, posts, threads, comments, transcripts, notes, drafts, manifestos, explainers
Existing PHILOSOPHY.md (optional but preferred for refinement)
Primary use context (recommended) — essays, political writing, technical criticism, cultural commentary, mixed use, etc.
Target use (optional) — AI prompting, editorial guidance, worldview mapping, archive/reference
Known convictions (optional) — beliefs the user already claims as core
Known anti-positions (optional) — positions, vibes, or frameworks the user rejects

If the user gives a giant dump with little explanation, proceed anyway. State assumptions and start with a corpus audit.

Corpus Quality Gates

Before extracting, assess the source material.

1. Source Type Weighting

Weight sources roughly like this:

casual but genuine writing
- comments
- transcripts
- notes
- rough drafts
long-form essays / posts
polished articles
commissioned copy / heavily constrained writing / debate performances aimed at an external frame

Highly edited or strategically constrained material is still useful, but it should not dominate the conclusions.

2. Exclusions

Exclude or downweight:

platform formatting tics
SEO filler
quotes from other people presented without endorsement
arguments made only to answer a narrow prompt and not seen elsewhere
obvious experiments that do not recur
ironic posturing with no supporting pattern

3. Sample Sufficiency

If the material is thin, say so clearly.

under ~500 words: too thin for a reliable worldview guide; produce only a provisional snapshot
500–1500 words: workable but limited
1500+ words across multiple pieces: solid
large corpus: ideal for full distillation

Always report confidence honestly.

Distillation Workflow

Follow this order.

Phase 1: Corpus Audit

Output a short assessment:

how much material exists
what kinds of material it is
which samples are strongest for authentic convictions
what will be excluded or downweighted
whether this is a first-pass distillation or refinement pass

Phase 2: Stable Pattern Extraction

Extract patterns that recur across the corpus:

primary use context cues
central questions
recurring themes
moral instincts
sacred values
recurring enemies / targets / irritants
favored explanations
assumptions about human nature
assumptions about power, institutions, truth, agency, obligation, freedom, dignity, beauty, progress, or decline
what tradeoffs the writer repeatedly chooses
where emotional intensity spikes and why
where reasoning quality repeatedly drops
where the writer goes vague, performative, evasive, inflated, or prematurely certain
which thinkers, traditions, schools, or concept families the writing most plausibly resonates with

Quote exact evidence where possible.

Where genuinely clarifying, map the worldview to existing intellectual frames such as:

thinkers
schools of thought
traditions
ideological lineages
named concepts

Examples of the type of move:

“closest to a disappointed Locke with anarchist tendencies”
“more Simone Weil in moral mood than Marxist in method”
“civilizationally conservative, but psychologically closer to existentialism than Burkean comfort”

Use these mappings to enrich the read, not to cosplay intelligence.

For primary use context, identify what kind of writing situation the philosophy seems optimized for:

reflective / essayistic
political / polemical / manifesto-like
analytical / explanatory
cultural criticism
strategic / persuasive
mixed

This is not mainly about audience demographics. It is about which intellectual and moral signals matter most in practice.

For delivery mode, identify whether the philosophy is usually carried through:

reflective / exploratory
declarative / thesis-driven
combative / oppositional
constructive / prescriptive
diagnostic / analytical
hybrid

This helps define how conviction is expressed without collapsing into generic writing advice.

Phase 3: Separate Stable Philosophy from Context Noise

For each pattern, decide whether it is:

canonical — strong recurring signal across contexts
probable — appears often, but not everywhere
tentative — interesting, but weak or context-bound

Never promote a weak signal to a hard rule just because it sounds profound.

Also separate:

core belief — the thing the writer actually seems to mean
habitual overreach — the place where they claim too much
evasion pattern — the place where they stop saying the sharp thing plainly
borrowed intensity — the place where rhetoric outruns substance

Phase 4: Draft or Refine `PHILOSOPHY.md`

If no PHILOSOPHY.md exists:

create one from the template in this skill

If a PHILOSOPHY.md exists:

preserve stable sections
sharpen vague claims
merge duplicates
remove unsupported rules
add stronger evidence
demote overconfident claims to tentative ones when needed

Phase 5: Crystallization

Create a top-of-document compression layer that can act as both summary and vibe check.

Include:

Philosophy Crystal — 1–3 lines capturing the worldview in compressed, memorable form
Failure Crystal — 1–2 lines capturing the most recurring self-deception or coherence failure
Thinker / Tradition Anchors — closest-fit references to existing thinkers, schools, or concepts

Rules for crystallization:

compression must be earned by the evidence, not just sound cool
prefer specificity over generic elegance
use existing thinkers or traditions when they genuinely clarify the pattern
if a thinker comparison is only partial, say so
if no single anchor fits, give 2–3 partial anchors instead of forcing a fake synthesis
if evidence is too thin, produce candidate crystal lines rather than pretending certainty

Phase 6: Validation

Generate a small validation pack:

yes, that tracks summary of the strongest commitments
doesn’t track contrast claims the corpus does not support
bullshit detector summary of the most recurring failure modes
crystal check whether the compression lines feel true, sharp, and memorable
what changed in the latest refinement pass
open questions where confidence is still low

Then ask for correction:

“What feels dead-on, what feels fake, and what feels overclaimed?”

Phase 7: Iterative Refinement Loop

When the user adds more writing or feedback, do targeted updates, not a full rewrite.

For each new pass:

compare new material against current PHILOSOPHY.md
confirm what strengthens existing claims
note what contradicts them
revise only the affected sections
append a short update note

Keep going until returns diminish.

Evidence Rules

These rules keep the skill from making shit up.

A. Every strong claim needs evidence

Whenever possible, support claims with:

repeated examples
quoted phrases
recurring judgments
cross-context recurrence

The same rule applies to negative judgments about the writer's reasoning. If you say they often dodge, overclaim, or hide behind abstraction, point to repeated evidence.

The same rule also applies to thinker or tradition comparisons. If you compare the writer to Locke, Weil, Nietzsche, Marx, Ellul, Lasch, or whoever else, the fit should be traceable to recurring patterns, not just aesthetic vibes.

B. Do not confuse topic with philosophy

If the user writes about politics, technology, religion, culture, gender, or institutions, do not assume the philosophy is just those topic labels. Look for the judgments, values, hierarchies, fears, loyalties, and preferred explanations that keep returning.

C. Do not confuse emotional intensity with core belief

A dramatic rant may be memorable without being central. Repetition beats drama.

D. Track recurring bullshit patterns without turning into a scold

Look for patterns like:

vague abstractions where a concrete claim should exist
complexity language used to avoid commitment
certainty performed before argument is earned
rhetorical force replacing reasoning
selective nuance used only when a claim becomes inconvenient
favorite conclusions reached faster than the evidence warrants

Name the pattern cleanly. Do not moralize or diagnose the soul.

E. Preserve contradictions when they are real

Some writers are both:

elitist and populist
skeptical and idealistic
anti-institutional and hungry for order
compassionate and punitive

Good philosophy docs preserve that tension instead of flattening it.

F. Distinguish diagnostic claims from normative claims

Separate:

what the writer thinks is true
what the writer thinks is good
what the writer thinks should be done

Do not blur those together.

G. Distinguish incoherence from development

Not every contradiction is hypocrisy or bullshit.

Sometimes the writer is:

genuinely changing their mind
testing a frame
speaking differently across contexts for legitimate reasons

Only call it a bullshit pattern when the evasion or overreach itself recurs.

H. Use thinker references as compression tools, not as substitute reasoning

Good:

“closest to civic republicanism with anti-bureaucratic instincts”
“moral psychology closer to Weil than to utilitarian reformism”

Bad:

random philosopher name-dropping with no explanatory gain
forcing a neat lineage where the corpus is obviously mixed
pretending the writer has read or endorsed the thinker being used as an anchor

What the `PHILOSOPHY.md` Should Include

Use the bundled template as the base structure.

At minimum, the output document should contain:

Core Orientation
- one-paragraph summary of the worldview
- default moral/intellectual posture
- what this writer seems fundamentally for and against
Crystal Layer
- a top-of-file compression of the worldview
- a top-of-file compression of the failure mode
- closest thinker / tradition / concept anchors
Primary Use Context
- the main writing situation this philosophy is optimized for
- what signals matter most in that context
Delivery Mode
- whether the philosophy is reflective, declarative, combative, constructive, diagnostic, or hybrid
- how that affects force, clarity, and interpretive risk
Central Questions
- what questions the writer keeps circling
Core Commitments
- recurring beliefs and values
Moral Instincts
- how the writer tends to judge actions, people, systems, and tradeoffs
Recurring Themes & Obsessions
- what topics or conflicts keep returning and why they matter
Targets, Enemies, and Resistances
- what the writer repeatedly pushes against
Preferred Explanations
- favored causal stories, lenses, and interpretive habits
Human Nature & Society Assumptions
- what the writer seems to assume about people, power, institutions, and collective life
Tensions & Contradictions
- genuine internal tensions worth preserving
Bullshit Detector
- recurring weak moves, evasions, overclaims, and coherence failures
Confidence & Scope
- where the writer sounds certain, exploratory, or conflicted
Evidence Bank
- exact excerpts or paraphrased evidence
- not just vibes
Open Questions
- unresolved tensions or unclear signals
Refinement Log
- what changed in each pass
AI Prompt Block
- compact prompt-ready summary for reuse

Required Behaviors

When creating the first draft

build a usable PHILOSOPHY.md, not a loose summary
include confidence labels where uncertainty exists
prefer a compact, readable artifact over bloated theory
favor definitions that are clear, non-contradictory where possible, and free of fake-smart buzzwords
include at least one section that names the writer's recurring bullshit patterns in a way that is precise, evidenced, and actually usable
include a top crystal layer that is memorable enough to serve as a quick vibe check later
prefer existing thinker / tradition / concept anchors when they genuinely enrich clarity

When refining an existing `PHILOSOPHY.md`

do not rewrite sections that still hold up
highlight what changed and why
remove claims that are not supported by the expanded corpus
improve evidence before inventing new categories

When evidence conflicts

preserve the contradiction if it appears genuine
otherwise split by context, such as:
- essays
- casual posts
- polemics
- explanatory writing

Output Format

After analysis, return:

a short corpus audit
the drafted or refined PHILOSOPHY.md
a concise refinement summary:
- added
- sharpened
- downgraded
- removed
- still uncertain
if applicable, a note on why the chosen thinker/tradition anchors fit better than nearby alternatives

Boundaries

This skill is great for:

distilling a worldview from many essays or posts
refining a living philosophy document over time
detecting recurring self-deception and argument failure patterns
preparing a reusable intellectual reference for AI or editors

This skill is not for:

diagnosing the writer’s personality or pathology
inventing a fake ideology from scratch
reducing a worldview to topic tags
doing lazy gotcha criticism from one edgy passage
deriving hard rules from tiny sample sets and pretending certainty

If the user wants actual writing or argument generation after the philosophy is distilled, use the resulting PHILOSOPHY.md as the input contract for preserving intellectual center.