@human-avatar/skills-for-humanity 1.0.0

package/skills/writing-argument/SKILL.md

@@ -0,0 +1,79 @@
+---
+name: writing-argument
+description: "Builds and repairs persuasive arguments by surfacing the warrant, auditing evidence, addressing counterarguments, and identifying rhetorical substitutes. Use when an argument has holes, the case isn't landing, or evidence isn't connecting to the claim. Triggers: 'build an argument', 'argument structure', 'the argument isn't working', 'op-ed', 'persuasive writing', 'make the case for', 'my argument has holes', 'the evidence doesn't land'."
+---
+
+# Writing: Argument
+
+Arguments fail not in the evidence but in the warrant — the unstated principle connecting evidence to claim. The warrant is the assumption that makes the argument work. Most writers never surface it, because to them it is obvious. But the reader may be supplying a different warrant — and reaching a different conclusion from the same evidence.
+
+The structure of every argument:
+- **Claim:** What you are asserting to be true
+- **Evidence:** The facts, data, examples, or observations that support it
+- **Warrant:** The principle or assumption that connects the evidence to the claim
+
+Example: "Remote workers are more productive (evidence). Therefore we should adopt a remote work policy (claim)." The warrant — "policies should be adopted when they increase productivity" — sounds obvious but is not. A reader who believes "productivity must be balanced against team cohesion and culture" will not accept the warrant, and the argument fails for them even if the evidence is strong. Making the warrant explicit forces both writer and reader to examine the actual assumption.
+
+The second most common failure: the strongest counterargument is not addressed. An argument that doesn't engage its best objection is not a complete argument — it is a one-sided brief that any reader with reservations can dismiss. Addressing the counterargument, not to concede but to answer it, is what makes an argument persuasive rather than merely insistent.
+
+---
+
+## Your Process
+
+**Step 1: Central Claim**
+State the argument's claim in one sentence. Be specific: "We should adopt a four-day work week" is more specific than "Work-life balance matters." The more specific the claim, the more clearly the evidence can be evaluated against it. A claim that can't be stated specifically is often a claim that hasn't been formed yet.
+
+**Step 2: Warrant — State It Explicitly**
+What principle or assumption connects the evidence to the claim? State it as a full sentence: "X is true (evidence), and because [warrant], therefore Y (claim)." Test: does the argument work if the reader doesn't share the warrant? If not, the warrant needs to be argued for — not just asserted.
+
+**Step 3: Evidence Audit**
+For each piece of evidence:
+- Is it sufficient? (Is there enough of it to support the claim, or is it a single data point generalised too broadly?)
+- Is it credible? (What is the source? Is it current? Is it directly relevant or analogous?)
+- Does it actually support the claim? (Sometimes evidence is relevant to the topic but doesn't logically connect to the specific claim being made)
+- Is anecdote being used where data is needed, or data being used where human experience would land harder?
+
+**Step 4: Counterargument**
+Identify the strongest objection to the claim — the best case that a reasonable, informed person in disagreement could make. Then assess: is this counterargument addressed in the piece? If it is not addressed, the argument has a hole that every skeptical reader will find. If it is addressed, is it addressed honestly (engaging the best version) or with a strawman (misrepresenting it to make it easy to defeat)?
+
+**Step 5: Rhetorical Substitutes**
+Flag appeals that are standing in for reasoning:
+- **Appeal to authority:** "Studies show..." without naming which ones; "experts agree..." — which experts? This is not evidence; it is the *claim* that evidence exists.
+- **Appeal to emotion:** Emotionally charged language or stories that produce a feeling of agreement without logical support — not wrong to use, but should not substitute for the argument.
+- **Repetition as intensification:** Restating the claim more forcefully as if repetition adds proof.
+
+These are not automatically illegitimate — emotion and authority are genuine persuasive tools. The problem is when they *substitute* for reasoning rather than support it.
+
+---
+
+## Output Format
+
+### Argument Audit
+
+**Central Claim:** [The argument in one sentence]
+
+**Warrant:** [Explicit statement of the connecting principle — acknowledged in the piece / reconstructed if implicit]
+
+**Evidence Quality:**
+- [Evidence piece] — Sufficient / Insufficient / Credible / Questionable / Relevant / Tangential
+- [Repeat]
+
+**Strongest Counterargument:** [The best objection to the claim] — Addressed / Not addressed / Strawmanned
+
+**Rhetorical Substitutes:**
+- [Quoted instance] — Type: [authority / emotion / repetition] — Substituting for: [what reasoning is missing]
+- NONE FOUND if clean
+
+**Verdict on Argument Strength:** [Strong / Has specific weaknesses / Weak — with diagnosis]
+
+**Reconstruction at Full Strength:** [The argument rebuilt — claim, warrant made explicit, evidence tightened, counterargument addressed, rhetorical substitutes replaced with reasoning]
+
+---
+
+## Notes
+
+- A reconstructed argument at full strength is the most useful output: it shows the writer what the argument would look like if it were complete. Often the reconstruction reveals that the warrant is actually contestable and needs its own supporting argument.
+- The most useful diagnostic question: "Why should I believe that this evidence implies this claim?" The answer is the warrant. If the answer is "isn't it obvious?", the warrant is unstated.
+- Pairs with `/writing-rhetoric` — rhetoric analysis examines how the argument is being made (its framing, its appeals, its assumptions); argument analysis examines whether the argument is sound.
+- Pairs with `/writing-restructure` — many argument problems are structural: the claim comes too late, the counterargument is buried, the warrant appears at the end when it needs to anchor the whole.
+- Pairs with `/writing-audience-calibration` — the warrant that works for one audience may not work for another; calibration includes choosing which assumptions to surface and argue for.

package/skills/writing-audience-calibration/SKILL.md

@@ -0,0 +1,72 @@
+---
+name: writing-audience-calibration
+description: "Calibrates writing for a specific reader by profiling their knowledge, concerns, and relationship to the topic — then rewriting for that reader without changing the substance. Use when content is too technical, too basic, or framed for the wrong audience. Triggers: 'write for a specific audience', 'audience calibration', 'this needs to work for non-technical readers', 'the audience isn't the right fit', 'calibrate this for', 'it's too technical / not technical enough', 'recalibrate this for'."
+---
+
+# Writing: Audience Calibration
+
+Calibration failures come in two forms: over-explanation and under-explanation. Over-explanation treats experts as novices — it defines terms they know, explains concepts they've mastered, and adds context they don't need. This reads as condescending, and the expert reader disengages. Under-explanation treats novices as experts — it uses jargon without definition, assumes mental models the reader doesn't have, and skips the connections that make the logic followable. This reads as inaccessible, and the novice reader gives up.
+
+The critical insight: calibration does not require changing the substance of what is being communicated. The same analysis can serve a technical expert and a non-technical decision-maker if it is correctly calibrated for each. The facts don't change; the entry point, assumed knowledge, vocabulary, framing, and emphasis all do.
+
+The three dimensions of calibration:
+- **Knowledge calibration:** What does this reader already know? What can be assumed, what needs brief context, what needs explanation?
+- **Stakes calibration:** What does this reader care about? The engineer cares about implementation; the product manager cares about user impact; the executive cares about business consequences. Same finding, different emphasis.
+- **Relationship calibration:** Is the reader expert or novice, friendly or skeptical, time-pressed or engaged? Each requires different structural choices.
+
+---
+
+## Your Process
+
+**Step 1: Reader Profile**
+Build a specific reader profile:
+- **Knowledge:** What domain knowledge, terminology, and conceptual background can be assumed?
+- **Role:** What is their function — technical, managerial, strategic? What decisions do they make?
+- **Stakes:** What do they care about most? What is the highest-value question they bring to this content?
+- **Relationship:** Friendly, skeptical, or neutral? Expert, novice, or intermediate?
+- **Time:** How much attention do they have? Will they read carefully or scan?
+
+**Step 2: Calibration Failures — Identify**
+Read the content as the profiled reader. Flag:
+- **Jargon without definition:** Technical terms the reader won't know
+- **Over-explanation:** Concepts or terms the reader already knows being explained to them
+- **Wrong prior assumed:** The content frames itself around a conceptual model the reader doesn't have
+- **Wrong emphasis:** The most important thing for *this reader* is buried; the most important thing for the *author* is prominent
+- **Wrong entry point:** The content starts in the middle of a context the reader is not inside
+
+**Step 3: Reader's Highest-Value Question**
+What is the single question this reader most needs answered? Is it answered prominently and early? If the reader has to read to the end to find the answer to their most pressing question, the calibration has failed regardless of other qualities.
+
+**Step 4: Framing Adjustment**
+Same substance, different framing. For technical content going to a non-technical reader: replace mechanism with outcome ("this algorithm sorts results by frequency" → "the most relevant results appear first"). For strategic content going to a technical reader: replace policy with specification ("improve response times" → "p95 response time under 200ms"). The framing makes the content land in the register the reader occupies.
+
+---
+
+## Output Format
+
+### Calibration Report
+
+**Reader Profile:** [Knowledge / Role / Stakes / Relationship / Time-attention]
+
+**Calibration Failures:**
+- [Jargon: quoted term + plain language alternative]
+- [Over-explanation: quoted passage + note on what the reader already knows]
+- [Wrong prior: identified assumption + what needs to be established instead]
+- [Wrong emphasis: what's prominent vs. what this reader needs prominent]
+- NONE FOUND if well-calibrated
+
+**Highest-Value Question:** [What this reader most needs answered] — Answered prominently / Buried / Not answered
+
+**Framing Adjustments:** [Specific reframings for this reader — mechanism → outcome, policy → specification, etc.]
+
+**Rewritten Opening Paragraph:** [The opening paragraph rewritten for the target reader — calibrated vocabulary, assumed knowledge, entry point, emphasis]
+
+---
+
+## Notes
+
+- Calibration is not simplification. A well-calibrated piece for a non-technical reader is not a dumbed-down version of the technical piece — it is the same substance approached from a different entry point. Simplification removes content; calibration changes the door.
+- The hardest calibration: from expert to novice, when the expert has forgotten what it's like not to know. The test: can a reader with the stated knowledge profile understand this without asking a clarifying question?
+- Pairs with `/writing-executive-summary` for executive calibration — the executive summary format is a specific calibration for senior decision-makers.
+- Pairs with `/writing-technical` for technical audience calibration — when the content is documentation and the calibration question is which technical level to write for.
+- Pairs with `/writing-voice-consistency` — calibration adjustments must stay within the established voice; a recalibrated piece that sounds like a different writer has traded one problem for another.

package/skills/writing-character-development/SKILL.md

@@ -0,0 +1,72 @@
+---
+name: writing-character-development
+description: "Engineers psychologically compelling characters by mapping want vs. need, wound, defence mechanism, and defining contradiction. Use when a character feels flat, underdeveloped, or unconvincing. Triggers: 'my character feels flat', 'character development', 'make this character feel real', 'this character isn't working', 'character work', 'how do I develop this character'."
+---
+
+# Writing: Character Development
+
+Flat characters come from describing traits rather than engineering contradictions. When a writer says "she's brave and determined," they have a label — not a person. Compelling characters feel real because they hold two things in tension simultaneously: they want something and need something different; they have a strength that doubles as a weakness; they act from a wound they can't fully see. These tensions generate unpredictable, specific behaviour that no list of adjectives could produce.
+
+The key diagnostic is the want/need split. The want is what the character consciously pursues — their stated goal, their driving motivation as they'd describe it. The need is what would actually fulfil them — which is always different, often the opposite, and usually something they're avoiding. This gap between want and need is where character lives. It generates internal conflict, drives the arc, and makes every action simultaneously understandable and complicated.
+
+---
+
+## Your Process
+
+**Step 1: State the Core Want vs. Core Need**
+Name what the character consciously wants (their goal as they understand it) and what they actually need (what would genuinely fulfil or redeem them). These should differ — the want is typically external, visible, and achievable; the need is internal, invisible to the character, and requires change. If want and need are identical, the character has no arc.
+
+**Step 2: Map the Wound**
+Identify the formative experience that created their worldview and defences. The wound is not a backstory fact — it's the event or pattern that taught them something about the world that now governs how they move through it. Name the belief the wound installed: "People leave if you show need," "Strength is the only thing others respect," "I am not enough." This belief is the character's operating system.
+
+**Step 3: Identify the Dominant Defence Mechanism**
+The wound generates a defence. Name the character's primary defensive pattern:
+- **Intellectualisation** — emotions managed through analysis; feelings translated into systems
+- **Avoidance** — proximity to the wound's trigger is avoided, often without full awareness
+- **Aggression** — threat is pre-empted by attacking first
+- **People-pleasing** — safety maintained by managing others' approval
+- **Controlling behaviour** — anxiety managed by controlling environment or others
+- **Dissociation / performance** — the self is presented as a crafted persona to keep the real self safe
+
+The defence should be visible in the character's behaviour patterns — how they handle conflict, intimacy, failure, and threat.
+
+**Step 4: Map Their Relationship to the Story's Central Tension**
+Is the character equipped or unequipped to handle the story's central challenge? The most powerful character-story alignment: the character's wound and defence are precisely the wrong tools for what the story requires. The story forces the character into exactly the territory their defence is designed to avoid.
+
+**Step 5: Define the Arc Endpoint**
+What must change internally for the character to reach the story's resolution? Not what they achieve externally — what must shift in their belief, their defence, their relationship to their wound? The arc endpoint should be the direct answer to the wound: if the wound installed "I am not enough," the arc endpoint might be "I am enough without proof."
+
+**Step 6: Define the Contradiction**
+Every compelling character holds two things in tension simultaneously — not one after the other, but at the same time. State the character's defining contradiction in one sentence. Examples: "Genuinely compassionate, uses that compassion as a weapon." "Desperate for connection, terrifies anyone who gets close." "The most honest person in the room, built on a foundational lie about themselves."
+
+---
+
+## Output Format
+
+### Character Profile
+
+**Core Want:** [What the character consciously pursues]
+
+**Core Need:** [What would actually fulfil them — what they're avoiding]
+
+**The Wound:** [The formative experience and the belief it installed]
+
+**Defence Mechanism:** [Primary pattern + how it manifests behaviourally]
+
+**Story Relationship:** [Equipped or unequipped to handle central tension, and why]
+
+**Arc Endpoint:** [What must internally change for resolution]
+
+**Defining Contradiction:** [The tension they hold simultaneously — one sentence]
+
+**Behavioural Tells:** [3–5 specific, observable behaviours that express the above — concrete enough to write on the page]
+
+---
+
+## Notes
+
+- Trait lists and adjectives produce flat characters. The wound, the defence, and the contradiction produce specific, unpredictable, recognisable people.
+- The behavioural tells are the most important output for the practitioner: they give you something you can actually put in the scene.
+- Pairs with `/writing-arc-design` when you need to map how this character's internal change aligns with the plot's external structure.
+- Pairs with `/writing-inconsistency-audit` when a character is behaving in ways that violate their established psychology — often visible as "they would never do that" moments.
+- Pairs with `/writing-dialogue` because voice flows from character: the wound and defence determine how a character speaks, evades, and misrepresents.

package/skills/writing-copy/SKILL.md

@@ -0,0 +1,83 @@
+---
+name: writing-copy
+description: "Writes and audits marketing copy, landing pages, ad copy, email copy, and product descriptions using the attention-desire-action framework. Use when copy isn't converting, describes features instead of benefits, or fails to earn the reader's next step. Triggers: 'copywriting', 'ad copy', 'landing page', 'email copy', 'marketing copy', 'write copy', 'the copy isn't converting', 'product description', 'headline isn't working'."
+---
+
+# Writing: Copy
+
+Copy fails when it describes the product rather than the reader's experience of having it. "Comprehensive workflow management platform with real-time collaboration" describes features. "Your team stops losing work in email chains" describes an experience. The reader buys the experience, not the feature set. Every piece of copy that describes the product from the product's point of view rather than the reader's is leaving conversion on the table.
+
+The three-layer model maps every effective piece of copy:
+
+**Attention (the opening):** The sole function of the first line is to earn the second line. Not to establish credibility, explain the product, or summarise the offer — to earn the next sentence. Attention is lost in the first three seconds. The headline must name a specific benefit or problem, not a category. "Project management software" is a category. "Your team ships on time, every time" is a claim that addresses a specific pain.
+
+**Desire (the body):** The body must make the reader feel the gap between their current state and a better one. This is not enthusiasm or puffery — it is specificity. "Boost productivity" is enthusiasm. "Cut the time your team spends on status updates from 3 hours a week to 10 minutes" is specific and credible. The desire is created not by telling the reader the product is good, but by making them feel what it would be like to have the problem solved.
+
+**Action (the CTA):** The call to action must be singular (one action only), specific (not "learn more" but "start your free trial"), and low-friction (the language should make action feel easy and obvious, not like a commitment or a risk).
+
+---
+
+## Your Process
+
+**Step 1: Attention — Does the Opening Earn the Next Sentence?**
+Read only the headline and first line. Stop. Ask: does this give the reader a reason to continue? Specifically:
+- Does it name a specific benefit, problem, or claim — not a category?
+- Does it speak to the reader's experience, not the product's features?
+- Is there a hook — something surprising, specific, or resonant?
+
+If the headline is a category label ("Enterprise HR Software") or a generic claim ("The Best Solution for Your Business"), it has failed at attention.
+
+**Step 2: Desire — Does the Body Create the Gap?**
+Read the body copy. Ask:
+- Does it make the reader feel the difference between their current state and a better one?
+- Are the benefits specific and credible? (Numbers, specifics, outcomes — not adjectives)
+- Does it address the reader's actual fear or resistance, or does it assume enthusiasm?
+- Is the social proof (if present) specific and relevant?
+
+**Step 3: Action — Is the CTA Singular, Specific, and Low-Friction?**
+Evaluate the call to action:
+- Is there only one action being asked for? (Multiple CTAs split attention and reduce conversion)
+- Is the language specific? ("Start free trial" > "Learn more" > "Click here")
+- Does the language make action feel easy? ("Takes 30 seconds" / "No credit card required" / "Cancel any time")
+- Is the CTA visible and prominent — or buried?
+
+**Step 4: Feature/Benefit Confusion**
+Flag every statement that describes the product (feature) rather than the reader's experience of having the product (benefit). For each:
+- Quote the feature statement
+- Identify the underlying benefit (what does this feature do *for the reader*?)
+- Write the benefit version
+
+**Step 5: Overall Verdict and Headline Rewrite**
+State the copy's single most significant failure. Rewrite the headline based on the strongest benefit identified in the copy.
+
+---
+
+## Output Format
+
+### Copy Audit
+
+**Attention Assessment:** [Opening quoted] — [Verdict: earns next sentence / fails at attention] — [Specific diagnosis]
+
+**Desire Audit:** [Benefit specificity and credibility / Gap-creation assessment / Fear/resistance addressed or ignored]
+
+**Action Assessment:** [CTA quoted] — [Singular / specific / low-friction check] — [Friction language identified]
+
+**Feature/Benefit Confusion:**
+- [Quoted feature statement] → [Benefit rewrite]
+- [Repeat]
+
+**Overall Verdict:** [Primary failure in one sentence]
+
+**Rewritten Headline:** [New headline addressing the strongest identified benefit]
+
+**Rewritten CTA:** [If the CTA is failing — specific, low-friction alternative]
+
+---
+
+## Notes
+
+- The single highest-return copy edit: turn the headline from a category or feature description into a specific benefit or outcome. This is almost always the first fix.
+- "More" is not a benefit. "Faster" is not a benefit. Faster *doing what*, resulting in *what outcome*, for *this specific reader* — that is a benefit.
+- Pairs with `/writing-audience-calibration` — copy must speak to the specific reader; generic copy that speaks to everyone reaches no one.
+- Pairs with `/writing-argument` for long-form copy (sales pages, long emails) where the copy is structured as an argument that needs to be logically sound as well as emotionally resonant.
+- Pairs with `/writing-tone-alignment` when copy spans multiple formats (headline, body, email sequence) and the tone needs to be consistent across them.

package/skills/writing-dialogue/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: writing-dialogue
+description: "Diagnoses and repairs dialogue for subtext, voice differentiation, exposition, and forward momentum. Use when dialogue sounds wrong, on-the-nose, or when characters all sound the same. Triggers: 'the dialogue sounds wrong', 'dialogue feels on the nose', 'dialogue review', 'characters sound the same', 'dialogue fix', 'dialogue is too direct', 'the dialogue explains too much'."
+---
+
+# Writing: Dialogue
+
+Dialogue fails in two primary ways: it says exactly what it means (no subtext), or it explains things the characters already know (exposition dressed as conversation). Both failures produce the same effect on the reader — a sense of flatness, of the author's hand visible, of characters becoming mouthpieces rather than people.
+
+Real conversation is oblique. People in conflict rarely say "I'm angry with you." They talk about something else. The anger is present in what they say about the dishes, the schedule, the way someone looks. The subtext *is* the scene — the surface conversation is the vehicle for it. When dialogue is on the nose, the subtext and the surface are the same thing, which means there is no scene — only information delivery.
+
+Voice differentiation is the second major failure point. Every character has a distinct rhythm of thought, a vocabulary range shaped by their history, evasion patterns specific to their wound, and a relationship to silence. When characters sound interchangeable, the dialogue's only differentiator is the attribution ("he said / she said") — which means the characters, on the page, don't exist yet.
+
+---
+
+## Your Process
+
+**Step 1: Speaker Goals — Surface vs. Subtext**
+For each speaker in the exchange, identify:
+- **Surface goal:** What they appear to be talking about
+- **Subtext goal:** What they actually want from this exchange (what they won't say directly)
+
+These should be different. If surface goal and subtext goal are identical, the speaker is saying exactly what they mean — the subtext has collapsed into the text, and the scene has lost its tension.
+
+**Step 2: Voice Differentiation**
+Analyse each speaker's distinct voice characteristics:
+- **Rhythm:** Long periodic sentences or short declarative ones? Interrupted speech or complete thoughts?
+- **Vocabulary range:** Formal/educated diction, colloquial speech, domain-specific language?
+- **Evasion pattern:** How does this character deflect, dodge, or redirect when pressed?
+- **Relationship to silence:** Does this character fill silences, create them, use them as weapons?
+
+If the voices are interchangeable — if you could swap the speakers' lines without the scene changing — the characters have not yet been individualised.
+
+**Step 3: Exposition Check**
+Flag any lines where characters explain things they both already know, solely for the reader's benefit. This is the "As you know, Bob" problem: "As you know, we've been married for fifteen years and your father never approved of me." Neither character needs to be told this. It's only there for the reader — and the reader knows it.
+
+Exposition that is genuinely needed can be delivered through conflict (characters who have different versions of the same event), curiosity (a character who genuinely doesn't know something), or revelation (information emerging under pressure, not volunteered).
+
+**Step 4: Scene Function**
+What is the dialogue doing? Mark all that apply:
+- **Conflict:** Two opposing wants in direct tension
+- **Revelation:** Something new emerges for a character or reader
+- **Bonding:** Intimacy or connection established or tested
+- **Negotiation:** Characters working toward an arrangement
+- **Deflection:** A character managing a threat through misdirection
+
+If the dialogue serves no discernible function — if it is simply exchange — it should be cut or redirected.
+
+**Step 5: Forward Momentum**
+Does the dialogue advance the scene, or does it stall it? Diagnose the rate at which the scene's situation changes through the dialogue. If the first and last lines of the dialogue exchange are in the same state as each other, the dialogue is stalling.
+
+---
+
+## Output Format
+
+### Dialogue Analysis
+
+**Speaker Goals:**
+- [Speaker A]: Surface goal / Subtext goal
+- [Speaker B]: Surface goal / Subtext goal
+- FLAGGED: [any speaker whose surface and subtext goals are identical]
+
+**Voice Differentiation:**
+- [Speaker A]: Rhythm / Vocabulary / Evasion pattern / Silence relationship
+- [Speaker B]: Same
+- FLAGGED: [any voices that are interchangeable]
+
+**Exposition Flags:** [Quoted lines + why they're exposition + how to deliver the information legitimately]
+
+**Scene Function:** [Conflict / Revelation / Bonding / Negotiation / Deflection — mark all present / FLAG if none]
+
+**Forward Momentum:** [Does the scene advance through dialogue? Where does it stall?]
+
+**Line-by-Line Notes:** [Specific notes on strongest exchanges and weakest — quote and diagnose]
+
+**Rewrites for Flagged Lines:** [Specific rewrites that restore subtext, differentiate voice, or remove exposition]
+
+---
+
+## Notes
+
+- Subtext does not mean obscure. The reader should be able to feel what's underneath even if the characters don't say it. Clarity and subtext are not in tension.
+- The best dialogue rewrites keep the surface plausible while loading the subtext — the characters are still talking about something real, but the real thing is not what they say.
+- Pairs with `/writing-scene-construction` for diagnosing the dialogue within its scene context — dialogue problems are often scene-construction problems.
+- Pairs with `/writing-character-development` because voice flows from character: the wound and defence determine how a character speaks, evades, and what they're incapable of saying directly.
+- Pairs with `/writing-voice-consistency` when the issue is that a character's voice is inconsistent across the manuscript rather than within a single scene.

package/skills/writing-executive-summary/SKILL.md

@@ -0,0 +1,68 @@
+---
+name: writing-executive-summary
+description: "Produces executive summaries, 1-page briefs, and board-level documents by extracting the situation, key findings, implications, and recommendation — answer-first, one page maximum. Use when a lengthy document needs a senior-audience brief that enables a decision. Triggers: 'executive summary', 'summarise this for a senior audience', '1-page brief', 'TL;DR for leadership', 'distil this', 'board summary', 'write a brief'."
+---
+
+# Writing: Executive Summary
+
+Executive summaries fail when they summarise the document rather than answering the reader's question. A summary of a 40-page analysis is not what an executive needs. They need: what is the situation, what are the three things most important to know, what does it mean for their decision, and what should happen next. In that order. In one page.
+
+The fundamental principle: the executive is not the end reader of your analysis — they are a decision-maker who needs the *output* of your analysis in a form that enables action. The detail, the methodology, the full data — that stays in the main document. The executive summary gives them what they need to act without reading the full document.
+
+Four common failures:
+- **Summarising the process, not the answer:** "We conducted a comprehensive analysis of three market segments, reviewing 47 data sources over 8 weeks." The executive doesn't need to know how long it took — they need to know what it found.
+- **Findings without implications:** Stating what happened without stating what it means for the decision at hand.
+- **False balance:** Including minor findings to be thorough, diluting the signal with noise. Three things that matter are better than twelve things that range from critical to marginal.
+- **Passive language on the recommendation:** "It may be worth considering..." — the executive summary is where the recommendation is made, not hedged.
+
+---
+
+## Your Process
+
+**Step 1: Reader's Role and Actual Decision**
+Who exactly is reading this? What is the specific decision they need to make? Not "the board should be informed" but "the board is deciding whether to approve the $2M market expansion budget." This decision frames everything: what gets included, what gets cut, and how the recommendation is framed.
+
+**Step 2: Identify the Three Most Important Things**
+From the full document: what are the three findings, facts, or insights that most directly bear on the reader's decision? Not the most interesting to you — the most important to them. These become the three bullets in the Key Findings section. If there are more than three, you are either including noise or you have not prioritised.
+
+**Step 3: Implications Not Findings**
+Translate findings into implications: not "our Q3 customer acquisition cost rose 40%" but "our current acquisition strategy is unsustainable at scale — at Q3 rates, the expansion budget funds half the projected customer growth." Implications answer "so what?" They are what the executive needs in order to act.
+
+**Step 4: Cut Context That Serves the Writer**
+Anything that explains how the analysis was done, why you looked at what you looked at, or what you ruled out — this serves the writer's need for credit and completeness, not the reader's need for decision support. Cut it. If it is genuinely important context, it can be one sentence.
+
+**Step 5: Draft to One Page / 300 Words**
+Every sentence must earn its place. The test for each sentence: does a reader without this sentence make a worse decision? If no, cut it. The executive summary is not a place for comprehensive coverage — it is a place for decisive clarity.
+
+---
+
+## Output Format
+
+### Executive Summary
+
+**Situation:** [One sentence. What is the context or problem that prompted this analysis?]
+
+**Key Findings:**
+- [Finding 1 — most important to the reader's decision]
+- [Finding 2]
+- [Finding 3 — maximum]
+
+**Implications:** [2–3 sentences. What do these findings mean for the decision? What is at stake if action is taken, or not taken?]
+
+**Recommendation:** [One clear sentence. What should happen? Active voice, no hedging.]
+
+**Next Steps:**
+- [Specific action, owner if relevant, timeline]
+- [Second action]
+- [Third action if needed]
+
+---
+
+## Notes
+
+- Max one page / 300 words. This is not a guideline — it is the requirement. An executive summary that requires two pages has failed.
+- The Situation sentence must be answer-first: it should be the one sentence that, if the executive reads nothing else, leaves them with the most important thing to know.
+- Passive language on recommendations ("it may be worth considering") signals that the writer is not confident in the recommendation. If you have done the analysis, make the recommendation. The executive can disagree — but they need a recommendation to react to.
+- Pairs with `/writing-report` — the executive summary sits above the full report. The report is for the reader who needs the detail; the summary is for the reader who needs the decision.
+- Pairs with `/writing-audience-calibration` — the executive's specific role, decision context, and prior knowledge shape every word of the summary; what an investor needs is different from what a CEO needs.
+- Pairs with `/writing-restructure` — if the source document is poorly structured, finding the three most important findings requires first identifying what the document actually says (which may require restructuring it).

package/skills/writing-inconsistency-audit/SKILL.md

@@ -0,0 +1,94 @@
+---
+name: writing-inconsistency-audit
+description: "Runs four systematic passes to identify timeline errors, character logic violations, world-rule breaks, and physical continuity errors. Use when a manuscript has contradictions, continuity problems, or logic violations. Triggers: 'there are contradictions', 'continuity errors', 'inconsistency check', 'the character shouldn't know that', 'timeline doesn't add up', 'continuity audit', 'inconsistencies in the manuscript'."
+---
+
+# Writing: Inconsistency Audit
+
+Inconsistencies undermine the reader's trust more efficiently than almost any other flaw — because they signal that the author lost track of their own world. The reader's contract with a story is: I will suspend disbelief in exchange for internal coherence. One visible inconsistency tells the reader the author may have lost the thread, and the reader will spend the rest of the story half-watching for the next crack rather than being inside the experience.
+
+Four types of inconsistency require separate passes, because they operate at different levels and have different causes:
+
+**Timeline inconsistency:** Events that are out of chronological order, time spans that don't add up, characters who are in two places at once, seasons that shift unexpectedly.
+
+**Character logic inconsistency:** Characters who act contrary to their established psychology without earned cause, who know things they couldn't know, who forget things they must know, who respond to situations in ways their established temperament doesn't allow.
+
+**World-rules inconsistency:** The story's established physics, magic, technology, or social rules that function differently in different scenes — usually because the plot needs them to.
+
+**Physical continuity inconsistency:** Objects that change hands, locations, or appearance without explanation; characters whose physical description changes; room layouts that contradict themselves.
+
+Each pass requires a different diagnostic lens. Running all four in sequence is the only way to catch inconsistencies that don't show up in a single read.
+
+---
+
+## Your Process
+
+**Pass 1: Timeline Audit**
+Reconstruct the chronological order of events. Note every time marker: "the next morning," "three weeks later," "before she left," "the day after Marcus died." Build a timeline and check:
+- Are events in order consistent with stated time markers?
+- Do stated durations (travel times, healing times, gestation times) match what the story shows?
+- Can characters be where they need to be at the times required?
+- Are ages consistent with established birthdates and story duration?
+
+Flag: any event whose placement contradicts a stated time marker; any duration that doesn't hold up to scrutiny.
+
+**Pass 2: Character Logic Audit**
+For each significant character, establish their baseline: what do they know, what is their temperament, what are they capable of? Then audit:
+- Does any character act on knowledge they couldn't have acquired by this point in the story?
+- Does any character act contrary to their established psychology without a scene that earns the change?
+- Does any character forget established facts (their own history, another character's name, an agreement they made)?
+- Does a character's relationship dynamic with another character contradict what was established earlier?
+
+Flag: each instance with the character name, the violated baseline, and the location.
+
+**Pass 3: World-Rules Audit**
+List the established rules of the world (physical, social, magical, technological). Then audit each instance where those rules appear:
+- Do the rules operate consistently throughout?
+- Are there moments where the rules bend because the plot needs them to?
+- Do characters know about the rules consistently with when they were established?
+- Are there loopholes the story exploits that the rules don't actually permit?
+
+Flag: each rule violation with the rule stated, the violation quoted, and whether it's a minor or structural inconsistency.
+
+**Pass 4: Physical Continuity Audit**
+Track key objects, locations, and physical descriptions:
+- Key objects: do they stay where they were left, appear when needed, disappear when forgotten?
+- Locations: do room layouts, relative positions, and environmental details remain consistent?
+- Physical descriptions: do characters' appearances remain consistent? (Hair colour, injuries, distinguishing features)
+- Scene-to-scene physical state: what does each character have, wear, and carry at the start of each scene? Does it match the end of the previous scene?
+
+Flag: each contradiction with the original description, the contradicting description, and the locations of both.
+
+---
+
+## Output Format
+
+### Inconsistency Report
+
+**Timeline:**
+- [Issue: quoted passage + location] — Severity: Minor / Breaks immersion / Breaks the story
+- NONE FOUND if clean
+
+**Character Logic:**
+- [Issue: character, baseline violated, quoted passage + location] — Severity
+- NONE FOUND if clean
+
+**World Rules:**
+- [Issue: rule stated, violation quoted + location] — Severity
+- NONE FOUND if clean
+
+**Physical Continuity:**
+- [Issue: original description, contradicting description, locations of both] — Severity
+- NONE FOUND if clean
+
+**Summary:** [Total issues by category and severity / The most critical fix required]
+
+---
+
+## Notes
+
+- Run the passes in order: timeline first, because timeline problems often explain character logic problems (a character knows something early because the scene was moved from later in the sequence).
+- Severity rating guide: **Minor** = a reader might notice but won't lose the story. **Breaks immersion** = pulls the reader out momentarily. **Breaks the story** = the story's central premise or resolution depends on an inconsistency.
+- Pairs with `/writing-character-development` for the character logic baseline — you need to know what was established before you can audit violations of it.
+- Pairs with `/writing-worldbuilding` for the world-rules baseline — the audit can only flag violations if the rules are clearly stated.
+- Pairs with `/writing-pov` because POV violations are a form of character logic inconsistency (the narration accesses knowledge the POV character doesn't have).