skilledagent 1.0.0

package/.agents/skills/writing-beats/SKILL.md

@@ -0,0 +1,67 @@
+---
+name: writing-beats
+description: Writing, exploit — assemble raw material into a journey of beats, grounding each term before a beat leans on it.
+disable-model-invocation: true
+---
+
+<what-to-do>
+
+The user has passed (or will pass) a markdown file of raw material. This is **exploit**: the exploring is done, the pile is fixed — commit to a path through it and mine the pile to fill each beat.
+
+If the user did not say where to save the article, ask once and remember the path.
+
+Then run a beat-by-beat journey, choose-your-own-adventure style:
+
+1. **Establish the prerequisites.** Before any beats, settle with the user what the audience already knows walking in — the concepts that are **grounded** from the start. Everything else must be grounded by a beat before a later beat can use it. See [Grounding](#grounding).
+2. Write 2–3 candidate **starting beats**, drawn from the raw material. Each is a different entry point into the article. Each may only lean on grounded concepts; note what new concepts each one grounds. Show the user the beats before writing to the article file. The user picks one. Preview what beats that pick unlocks — as if the user is seeing a little way down the path.
+3. Once the user picks a starting beat, write **only that beat** to the article file. A beat may be one sentence or several paragraphs — whatever that beat naturally is. Stop there.
+4. Re-read the article file from disk. Then offer 2–3 candidate **next beats** — different directions the journey could pivot to from where the article now stands. Each must be reachable from the current grounded set; note what each one grounds.
+5. Loop steps 3–5 until the article reaches a natural end.
+
+</what-to-do>
+
+<supporting-info>
+
+## Grounding
+
+Every **concept** has to be **grounded** before a beat can lean on it: the audience either walked in knowing it or met it in an earlier beat. A beat that reaches for an ungrounded concept loses the reader — that is the one move the journey can't make. The unit is the concept, not the word for it: a beat can lean on an idea the reader lacks even with no jargon in sight. Where a concept has a name — a **term** — grounding it means landing the idea and the term together.
+
+A concept gets grounded one of two ways:
+
+- **Prerequisite** — grounded before the first beat. The audience brings it. Fixed at the start.
+- **Introduced** — a beat establishes it, and from then on it's grounded for every later beat.
+
+So each beat does two jobs: it **requires** concepts that are already grounded, and it **grounds** new ones. Keep a running list of what's grounded so far, and update it each time a beat lands.
+
+This is what shapes the choose-your-own-adventure. A candidate beat is only reachable if everything it requires is already grounded; picking a beat that grounds concept X unlocks every beat that was waiting on X. When you offer next beats, they must all be reachable from the current grounded set — and say what each one grounds, so the user can see which paths it opens.
+
+The big lever is what you make a prerequisite versus what you ground inside the piece. Demand too much up front and you shut out readers who don't have it; ground too much inside and the early beats drown in definitions. Settle this with the user when you establish prerequisites, and revisit it whenever a tempting beat turns out to require a concept nothing has grounded yet — the fix is either a grounding beat before it, or promoting the concept to a prerequisite.
+
+## What is a beat
+
+A beat is one move in the journey. It does one thing — sets a scene, lands a point, asks a question, drops an aside, twists the angle. Then it stops, leaving the reader at a place where the next beat can pivot.
+
+A beat is sized by what it needs:
+
+- A single sentence if that's all the move is ("And then nothing happened for three weeks.").
+- A short paragraph if the move needs setup.
+- Multiple paragraphs if the beat is a self-contained vignette, argument, or example.
+
+If a "beat" needs five paragraphs and three subheadings, it's not a beat — it's two beats glued together. Split it.
+
+## Pulling from the pile
+
+Pull material from the raw pile to populate each beat. You can paraphrase, split, recombine, or quote. The pile is a quarry.
+
+## Ending the journey
+
+The article ends when the journey is complete — not when the pile is empty. Most piles will have leftover fragments that don't make it in. That is fine; that is the point of having more raw material than you need.
+
+## Writing rhythm
+
+- Append one beat at a time. Never write ahead.
+- Re-read the article file from disk before every write. Preserve user edits absolutely.
+- If the user edits a previous beat substantially, let it change what comes next.
+- If the user says "rewrite that beat" or "go back and try a different beat 3", do it — edit in place, leave the rest alone.
+
+</supporting-info>

package/.agents/skills/writing-fragments/SKILL.md

@@ -0,0 +1,79 @@
+---
+name: writing-fragments
+description: Writing, explore — mine raw fragments, no structure yet.
+disable-model-invocation: true
+---
+
+<what-to-do>
+
+This is pure **explore**: widen the space of what could be written without committing to structure — committing is _exploit_, a separate skill's job. Run a grilling session that produces fragments, interviewing the user relentlessly about whatever they want to write about. Imposing phases, outlines, or article structure is out of scope here.
+
+As fragments emerge from either side of the conversation, append them to a single markdown file.
+
+If the user did not pass a path, ask once where to save the document, then remember it for the rest of the session.
+
+Capture fragments from the very first thing the user says, including the initial prompt.
+
+On first write, put a single H1 at the top with a working title (it can change later) and nothing else — no metadata, no TOC, no date.
+
+</what-to-do>
+
+<supporting-info>
+
+## What is a fragment
+
+A fragment is any piece of text that might survive into the final article. It must be _readable by the author_ — the author can tell what it means — but it does not need to define its terms or be comprehensible to a cold reader. The bar is "is this a piece of good writing?", not "is this a self-contained argument?"
+
+Fragments are deliberately heterogeneous. Examples of what could be a fragment:
+
+- A sharp sentence you'd want to deploy somewhere but don't yet know where.
+- A claim with a one-line justification.
+- A vignette: a thing that happened, a code snippet, a scenario, an analogy.
+- A half-thought: "something about how X feels like Y, work this out later."
+- A quote, a piece of dialogue, an overheard line.
+- A list of related observations that hang together by feel.
+- A complaint, a confession, a punchline.
+- A **leading word** — a compact metaphor or coinage the whole piece can hang on (one term that names the idea, the way _tracer bullets_ or _fog of war_ names a whole pattern).
+
+Of these, the leading word is the most valuable fragment to land. It is load-bearing: name the right one in explore and it shapes the structure, the transitions, and the title later — paying dividends through the entire exploit phase. When the conversation circles a recurring idea, push to coin a word for it.
+
+The novelist's diary is the model: years of unstructured noticings that later get mined for raw material. Fragments are noticings.
+
+## File format
+
+```markdown
+# Working title
+
+A first fragment lives here.
+
+It can be multiple paragraphs. It can include lists, code, quotes — whatever
+shape the fragment naturally takes.
+
+---
+
+A second fragment.
+
+---
+
+> A quoted line that the user wants to keep around.
+
+A reaction to it.
+
+---
+
+- A cluster of related observations
+- That hang together by feel
+- And want to be near each other
+```
+
+Fragments are separated by a horizontal rule (`\n---\n`). No headings inside the body. No tags. No order beyond the order they were added.
+
+## Writing rhythm
+
+Append silently. Don't ask permission for each fragment. Mention what you added in passing ("adding that"), but don't interrupt the conversation with save dialogs.
+
+Before every write: re-read the file from disk. The user may have edited, reordered, or deleted fragments between turns — preserve their changes. Never overwrite the file; only append (or, if the user asks, edit a specific fragment in place).
+
+The user can say "cut the last one", "rewrite that one sharper", "merge those two" at any time. Treat those as first-class instructions.
+
+</supporting-info>

package/.agents/skills/writing-great-skills/GLOSSARY.md

@@ -0,0 +1,201 @@
+# Glossary — Building Great Skills
+
+The domain model for what makes a skill great. A skill exists to wrangle determinism out of a stochastic system; the root virtue is **Predictability**, and every term below is a lever on it. This is the disclosed reference for [`writing-great-skills`](SKILL.md).
+
+The terms are grouped by axis: **Invocation** (how a skill is reached), **Information Hierarchy** (how its content is arranged), **Steering** (how the agent's runtime behaviour is shaped), and **Pruning** (how it is kept lean). Each **failure mode** lives beside the lever that cures it, tagged _failure mode_.
+
+**Bold terms** in any definition are themselves defined in this glossary; find them by their heading.
+
+## Predictability
+
+The degree to which a skill makes the agent behave the same _way_ on every run — the same process, not the same output (a brainstorming skill should _predictably_ diverge; its tokens vary, its behaviour doesn't). The root virtue every other term serves — cost and maintainability are symptoms of it, not rivals.
+
+_Avoid_: consistency, reliability, robustness, output-determinism
+
+## Invocation
+
+How a skill is reached — and the two loads you pay for the choice.
+
+### Model-Invoked
+
+A skill that keeps its **description** field, so the agent can see it and fire it autonomously — and the human can still type its name, so model-invocation always _includes_ user reach. There is no model-only state: a description only ever _adds_ agent discovery, never removes the human's. Pays a permanent **context load** on every turn in exchange for that discoverability. Reachable by other skills, because the description that makes it agent-discoverable makes it invocable. A model-invoked skill whose content is all **reference** is also one home for shared reference: another skill can invoke it, so reference needed by several skills lives in one place. Pick model-invocation only when the agent must reach the skill on its own; if it never fires except by hand, drop the description and pay no context load.
+
+_Avoid_: ability, tool, capability
+
+### User-Invoked
+
+A skill with its **description** stripped — invisible to the agent and reachable only by the human typing its name (user-_only_, where **model-invoked** is user-_and-agent_). Trades agent-discoverability for zero **context load**. Because it has no description, nothing but the human can reach it: no other skill can fire it.
+
+_Avoid_: procedure, workflow, command
+
+### Description
+
+The skill's machine-readable trigger, and the one **context pointer** a **model-invoked** skill is forced to keep loaded at all times. Its mere presence _is_ the invocation axis: keep it and the skill is model-invoked (and reachable by other skills); delete it and the skill is **user-invoked**, reachable only by the human. The source of a model-invoked skill's **context load**.
+
+_Avoid_: frontmatter, summary
+
+### Context Pointer
+
+A reference held in the agent's context that names some out-of-context material and encodes the condition for reaching it. The **description** is the top-level context pointer (context window → skill); pointers to disclosed files are the same object one level down. Its wording, not the target, decides _when_ the agent reaches — and _how reliably_. A must-have target behind a weakly worded pointer is a variance bug: fix the wording first, and inline the material only if sharpening fails.
+
+_Avoid_: link, reference, import
+
+### Context Load
+
+The cost a **model-invoked** skill imposes on the agent's context window — its **description**, always loaded, spending both tokens and attention. What **user-invoked** skills escape by having no description, and the brake on splitting into more model-invoked skills.
+
+_Avoid_: token cost, context bloat
+
+### Cognitive Load
+
+The cost a **user-invoked** skill imposes on the human — what they must hold in their head: which skills exist and when to reach for each (the human is the index). What **model-invocation** removes by being agent-discoverable, and the brake on splitting into more user-invoked skills. Not a cost to minimise: it is the price of human agency, the reason some skills stay user-invoked. Spend it where human judgement matters; remove it where it does not.
+
+_Avoid_: human index, burden, overhead
+
+### Router Skill
+
+A **user-invoked** skill whose job is to point at your other user-invoked skills — naming each and when to reach for it — so the human has one skill to remember instead of many. It can only hint, never fire them: user-invoked skills have no **description**, so nothing but the human can reach them. The cure for **cognitive load** when user-invoked skills multiply.
+
+_Avoid_: dispatcher, menu, registry, index, router procedure
+
+### Granularity
+
+How finely you divide skills. Finer division spends one of the two loads: more **model-invoked** skills spend **context load** (more descriptions crowding the window and competing for attention); more **user-invoked** skills spend **cognitive load** (more for the human to remember and reach for). Two cuts guide the division. By **invocation**, split off a model-invoked skill where you have a distinct **leading word** to trigger it — a trigger word you actually use in your prompts. By **sequence**, split a run of **steps** where a step's **post-completion steps** need hiding, since isolating it in its own context clears what follows. Beware the reverse: merging sequences exposes each step's post-completion steps to what follows, inviting premature completion.
+
+_Avoid_: chunking, modularity
+
+## Information Hierarchy
+
+How a skill's content is arranged, and how far down the ladder each piece sits.
+
+### Information Hierarchy
+
+A skill's content ranked by how immediately the agent needs it — a single ladder, produced by two cuts: in-file or behind a pointer, and step or reference. The rungs:
+
+- **Steps** — in-file, primary
+- **Reference**, in-file — secondary
+- **Reference**, disclosed — behind a **context pointer**
+
+A skill with no **steps** uses just the bottom two rungs — often a legitimately flat peer-set (e.g. every rule of a review on one rung), which is a fine arrangement, not a smell. The hierarchy is independent of invocation: a skill can be model- or user-invoked whether it is all steps, all reference, or both. When a skill has steps, in-file reference that should be disclosed buries them and turns attending to them into a coin-flip — a variance lever, not just a legibility one. Keep the top of the ladder legible; push down it whatever you can.
+
+_Avoid_: structure, organization, layout
+
+### Steps
+
+The ordered actions the agent performs — when a skill has them, the primary tier of its content, and the part that earns its place in SKILL.md. Not every skill has steps: a skill can be all steps (`tdd`), all **reference** (a review), or both, independent of invocation. Every step ends on a **completion criterion**, clear or vague.
+
+_Avoid_: workflow, instructions, choreography
+
+### Reference
+
+Material the agent refers to on demand — definitions, facts, parameters, examples, conditional instructions. When a skill has **steps** it is secondary to them; when a skill has none it is the entire content; or it lives outside any skill entirely — see **External Reference**. Reached via **context pointers**, and the prime candidate for **progressive disclosure**.
+
+_Avoid_: supporting material, docs, background
+
+### External Reference
+
+**Reference** that lives outside the skill system — a plain file, no **description**, no **steps**, not invocable — that any skill can point at. The home for shared reference that needn't fire on its own, and the only shared home two **user-invoked** skills can use, since neither has a description and so neither can fire the other.
+
+_Avoid_: doc, resource, knowledge base
+
+### Progressive Disclosure
+
+Moving **reference** down the ladder — out of SKILL.md and behind a **context pointer** — so the top stays legible. Not primarily a token optimisation; it is how the **information hierarchy** is protected. Licensed by **branching**: disclose what only some branches need, inline what every path needs, and if a pointer fires unreliably on must-have material, sharpen its wording, and pull it back inline only if that fails.
+
+_Avoid_: lazy loading, chunking
+
+### Co-location
+
+Keeping the material an agent needs at once in one place — a concept's definition, rules, and caveats under a single heading, not scattered across the file — so reading one part brings its neighbours with it. The within-file companion to the **Information Hierarchy**: the hierarchy ranks _how far down_ a piece sits; co-location decides _what sits beside it_ once there. There is no formula for the right format of a body of **reference**; the test is that a skill should read like documentation written for the agent, and grouped material reads that way where scattered material does not. Distinct from **Duplication**: that repeats one meaning in two places, where scattering fragments a single meaning across many.
+
+_Avoid_: grouping, clustering, cohesion
+
+### Sprawl
+
+_Failure mode._ A skill that is simply too long — too many lines in SKILL.md — independent of whether they are stale or repeated. Even an all-live, all-unique skill can sprawl. It costs readability (the agent wades through more before it can act, and attention thins across the excess), maintainability (every extra line is one more to keep **relevant**), and tokens. The cure is the **information hierarchy**: push **reference** down behind **context pointers**, and split by **branch** or sequence so each path carries only what it needs. Distinct from **sediment** (length from stale accumulation) and **duplication** (length from repeated meaning) — sprawl is length itself, whatever its cause.
+
+_Avoid_: bloat, length, size, verbosity
+
+## Steering
+
+The levers that shape the agent's runtime behaviour toward **Predictability**.
+
+### Branch
+
+A distinct way a skill can be invoked — a case the skill handles — so different runs take different paths through it. A skill with many steps may carry many branches; a linear one has none.
+
+_Avoid_: path, case, fork
+
+### Leading Word
+
+A compact concept — also called a _Leitwort_ — already living in the model's pretraining, that the agent thinks with while running the skill. It encodes a behavioural principle in the fewest possible tokens by invoking priors the model already holds (e.g. _lesson_, _proximal zone of development_, _fog of war_, _tracer bullets_). Repeated as a token, never as a sentence, it accumulates a distributed definition across the skill and anchors a whole region of behaviour. Coining your own works if you define it clearly, but a made-up word recruits no priors — you pay in definition tokens what a pretrained word gives free. Reach for an existing word first.
+
+A leading word serves **predictability** twice. In the body it anchors **execution** — the agent reaches for the same behaviour every time the concept appears, and inside flat reference it focuses attention on a class of thing to look for, recruiting the right checks each run. In the **description** it anchors **invocation** — and not only within the skill: when the same word lives in your prompts, your docs, and your codebase, the agent links that shared language to the skill and fires it more reliably. Word a description with the leading words you actually use when you want the skill.
+
+_Avoid_: keyword, term, motif
+
+### Completion Criterion
+
+The condition that tells the agent a unit of work is done — the target it judges against. Two properties make it a lever, not just a quality. Its **clarity** (can the agent tell done from not-done?) resists **premature completion** — a vague bound ("understanding reached") lets the agent declare done and slip to the next step; this axis needs _steps_ to bite, since premature completion is a between-steps failure. Its **demand** (how much it requires) sets **legwork** — "every modified model accounted for" forces thorough work where "produce a change list" does not — and this axis is _not_ step-bound: it can bind a body of flat reference too, which is how a skill with no steps still carries an exhaustiveness bar ("every rule applied"). The strongest criteria are both checkable and exhaustive.
+
+_Avoid_: done condition, exit condition, stopping rule
+
+### Legwork
+
+The work an agent does behind the scenes within a single step — reading files, exploring the codebase, making changes, digging up what it needs rather than offloading to the user. It lives below the step structure: never written as its own step, latent in the wording, controlled by the agent rather than the skill. The within-step counterpart to **post-completion steps**' across-step pull. Raised by a **leading word** (_comprehensive_, _thorough_) or a **completion criterion** that demands the work be exhaustive — including the demand axis applied to flat reference, which is what drives a skill of flat reference to cover all its rungs. Goes thin either when that demand is missing or when **premature completion** cuts the step short.
+
+_Avoid_: scope, effort, diligence, coverage
+
+### Post-Completion Steps
+
+The **steps** that follow the current step. Visible, they pull the agent forward into **premature completion** — the more it sees, the stronger the tug; the defence is to hide them by splitting the sequence of steps into two.
+
+_Avoid_: horizon, fog of war, lookahead
+
+### Premature Completion
+
+_Failure mode._ Ending the current step before it is genuinely done, because the agent's attention slips to being done rather than to the work. A between-steps failure: it needs **steps** to occur — a skill with no steps that quits early isn't premature completion but thin **legwork** under an unmet demand. A tug-of-war between two forces: visible **post-completion steps** (the pull forward) and the **completion criterion**'s clarity (the resistance — a sharp, checkable bar holds; a vague one gives way). Fuzziness is the necessary condition: a sharp bound resists the pull no matter how many later steps are visible, so a step that never rushes needs no defending. Two levers hold a step that does, but reach for them in order: **sharpen the bound first** — it is local and cheap. Only when the criterion is irreducibly fuzzy _and_ you actually observe the rush do you **hide the later steps** — and hiding only works across a real context boundary (a user-invoked hand-off or a subagent dispatch; an inline model-invoked call leaves the later steps in context and clears nothing). One cause of thin legwork, but distinct from it: legwork can be thin even when a step runs to full completion.
+
+_Avoid_: premature closure, the rush, rushing, shortcutting
+
+### Negation
+
+_Failure mode._ Steering by prohibition — telling the agent what _not_ to do — which drags the forbidden behaviour into context and makes it _more_ available, not less. _Don't think of an elephant_, and the elephant is all there is; _never write verbose comments_, and verbosity is the pattern the agent has just read. The negation is a weak modifier the strongly-activated concept overruns, so the ban half-reads as an instruction to do the thing. Its **leading word** is the _elephant_: whatever a prohibition names into the frame. Cure: prompt the **positive** — describe the target behaviour ("write one-line comments") so the banned one is never spoken. A prohibition earns its place only as a hard guardrail on a behaviour you cannot phrase positively; even then, pair it with the positive target so attention lands on what to do.
+
+_Avoid_: ironic rebound, don't-prompting, the pink elephant
+
+## Pruning
+
+Keeping a skill lean — each remedy paired with the failure it cures.
+
+### Single Source of Truth
+
+The desired state where each meaning lives in exactly one authoritative place, so a change to the skill's behaviour is a change in one place. **Duplication** is its violation.
+
+_Avoid_: home, canonical location
+
+### Duplication
+
+_Failure mode._ The same meaning given more than one **single source of truth**. It costs maintenance (change one place, you must change the others), costs tokens, and inflates prominence — repeating a meaning weights it on the ladder past its real rank. The accidental inverse of a **leading word**, which raises attention on purpose by repeating a token, never the meaning.
+
+_Avoid_: repetition, redundancy
+
+### Relevance
+
+Whether a line still bears on what the skill does — the lens for what to keep. A line loses relevance either by never bearing on the task (mere exposition, or a **branch** that should be disclosed) or by going stale: drifting out of date as the behaviour or world it describes changes. Shorter skills are easier to keep relevant, because each line is cheaper to check. Distinct from **no-op**: relevance asks whether a line bears on the task, not whether it changes behaviour.
+
+_Avoid_: load-bearing, staleness, freshness
+
+### Sediment
+
+_Failure mode._ Layers of old content that settle in a skill and are never cleared, because adding feels safe and removing feels risky — so stale and irrelevant lines accumulate and you must core down through them to find what is still live. The default fate of any skill without a pruning discipline; the slow erosion of **relevance**, as opposed to **duplication**'s repeated meaning.
+
+_Avoid_: accretion, bloat, cruft, rot
+
+### No-Op
+
+_Failure mode._ An instruction that changes nothing because the model already does it by default — you pay load to tell the agent what it would do anyway. The test: does a line change behaviour versus the default? A line can be perfectly **relevant** and still be a no-op. The same priors that make a **leading word** free make a no-op worthless.
+
+A leading word is a _technique_; No-Op is a _verdict_ on a line — and they cross. A leading word too weak to beat the default is a no-op (_be thorough_ when the agent is already thorough-ish), and the fix is a stronger word that passes the verdict (_relentless_), not a different technique. So the No-Op test — does it change behaviour versus the default? — is also how you grade whether a leading word is earning its repetitions. This is model-relative, not reader-relative: two people disagreeing over whether a line is a no-op disagree about the default, and settle it by running the skill, not by debate.
+
+_Avoid_: redundant instruction, restating the obvious, belaboring

package/.agents/skills/writing-great-skills/SKILL.md

@@ -0,0 +1,83 @@
+---
+name: writing-great-skills
+description: Reference for writing and editing skills well — the vocabulary and principles that make a skill predictable.
+disable-model-invocation: true
+---
+
+A skill exists to wrangle determinism out of a stochastic system. **Predictability** — the agent taking the same _process_ every run, not producing the same output — is the root virtue; every lever below serves it.
+
+**Bold terms** are defined in [`GLOSSARY.md`](GLOSSARY.md); look them up there for the full meaning.
+
+## Invocation
+
+Two choices, trading different costs:
+
+- A **model-invoked** skill keeps a **description**, so the agent can fire it autonomously _and_ other skills can reach it (you can still type its name too). It contributes to **context load** — the description sits in the window every turn. Mechanics: omit `disable-model-invocation`, and write a model-facing description with rich trigger phrasing ("Use when the user wants…, mentions…").
+- A **user-invoked** skill strips the description from the agent's reach: only you, typing its name, can invoke it — and no other skill can. Zero context load, but it spends **cognitive load**: _you_ are the index that must remember it exists. Mechanics: set `disable-model-invocation: true`; the `description` becomes human-facing — a one-line summary, trigger lists stripped.
+
+Pick model-invocation only when the agent must reach the skill on its own, or another skill must. If it only ever fires by hand, make it user-invoked and pay no context load.
+
+When user-invoked skills multiply past what you can remember, that piled-up cognitive load is cured by a **router skill**: one user-invoked skill that names the others and when to reach for each.
+
+## Writing the description
+
+A model-invoked **description** does two jobs — state what the skill is, and list the **branches** that should trigger it. Every word increases **context load**, so a description earns even harder pruning than the body:
+
+- **Front-load the skill's leading word** — the description is where it does its invocation work.
+- **One trigger per branch.** Synonyms that rename a single branch are **duplication** — "build features using TDD … asks for test-first development" is one branch written twice. Collapse them; keep only genuinely distinct branches.
+- **Cut identity that's already in the body.** Keep the description to triggers, plus any "when another skill needs…" reach clause.
+
+## Information hierarchy
+
+A skill is built from two content types — **steps** and **reference** — that mix freely: a skill can be all steps, all reference, or both. The core decision is which to use and where each sits on the **information hierarchy**, a ladder ranked by how immediately the agent needs the material:
+
+1. **In-skill step** — an ordered action in `SKILL.md`, the primary tier: what the agent does, in order. Each step ends on a **completion criterion**, the condition that tells the agent the work is done. Make it _checkable_ (can the agent tell done from not-done?) and, where it matters, _exhaustive_ ("every modified model accounted for", not "produce a change list") — a vague criterion invites **premature completion**.
+2. **In-skill reference** — a definition, rule, or fact in `SKILL.md`, consulted on demand. Often a legitimately flat peer-set (every rule of a review on one rung) — a fine arrangement, not a smell. _This skill is all reference._
+3. **External reference** — reference pushed out of `SKILL.md` into a separate file, reached by a **context pointer**, loaded only when the pointer fires. (Spans _disclosed_ reference — a sibling file like `GLOSSARY.md`, still part of the skill — through fully **external reference** that lives outside the skill system and any skill can point at.)
+
+A demanding completion criterion drives thorough **legwork** — the digging the agent does within the work — whether the skill has steps or not, since "every rule applied" binds flat reference just as "every step done" binds a sequence.
+
+Push too little down and the top bloats; push too much and you hide material the agent actually needs. That tension is the whole decision.
+
+**Progressive disclosure** is the move down the ladder — out of `SKILL.md` into a linked file — so the top stays legible. Mechanics: a linked `.md` file in the skill folder, named for what it holds (this skill discloses its full definitions to `GLOSSARY.md`). Some skills are used in more than one way, and each distinct way is a **branch** — different runs taking different paths through the skill. Branching is the cleanest disclosure test: inline what every branch needs, and push behind a pointer what only some branches reach. A **context pointer**'s _wording_, not its target, decides when and how reliably the agent reaches the material.
+
+Where the ladder decides _how far down_ a piece sits, **co-location** decides _what sits beside it_ once there: keep a concept's definition, rules, and caveats under one heading rather than scattered, so reading one part brings its neighbours with it.
+
+## When to split
+
+**Granularity** is how finely you divide skills, and each cut spends one of the two loads, so split only when the cut earns it. Two cuts:
+
+- **By invocation** — split off a **model-invoked** skill when you have a distinct **leading word** that should trigger it on its own, or another skill must reach it. You pay **context load** for the new always-loaded **description**, so that independent reach has to be worth it.
+- **By sequence** — split a run of **steps** when the steps still ahead (a step's **post-completion steps**) tempt the agent to rush the one in front of it (**premature completion**). Keeping them out of view encourages the agent to do more **legwork** on the current task.
+
+## Pruning
+
+Keep each meaning in a **single source of truth**: one authoritative place, so changing the behaviour is a one-place edit.
+
+Check every line for **relevance**: does it still bear on what the skill does?
+
+Then hunt **no-ops** sentence by sentence, not just line by line: run the no-op test on each sentence in isolation, and when one fails, delete the whole sentence rather than trim words from it. Be aggressive — most prose that fails should go, not be rewritten.
+
+## Leading words
+
+A **leading word** is a compact concept already living in the model's pretraining that the agent thinks with while running the skill (e.g. _lesson_, _fog of war_, _tracer bullets_). Repeated throughout the text (though not necessarily - a strong leading word might only be needed once), it accumulates a distributed definition and anchors a whole region of behaviour in the fewest tokens, by recruiting priors the model already holds.
+
+It serves predictability twice. In the body it anchors _execution_: the agent reaches for the same behaviour every time the word appears. In the description it anchors _invocation_: when the same word lives in your prompts, docs, and code, the agent links that shared language to the skill and fires it more reliably.
+
+Hunt for opportunities to refactor skills to use leading words. A triad spelled out at three sites (**duplication**), a description spending a sentence to gesture at one idea — each is a passage begging to **collapse** into a single token. Examples include:
+
+- "fast, deterministic, low-overhead" -> _tight_ — one quality restated across a phase — into a single pretrained word (a _tight_ loop).
+- "a loop you believe in" -> _red_ — converts a fuzzy gate into a binary observable state (the loop goes _red_ on the bug, or it doesn't).
+
+You win twice over: fewer tokens, _and_ a sharper hook for the agent to hang its thinking on. Assume every skill is carrying restatements that leading words retire — go find them.
+
+## Failure modes
+
+Use these to diagnose issues the user may be having with the skill.
+
+- **Premature completion** — ending a step before it's genuinely done, attention slipping to _being done_. Defence, in order: sharpen the completion criterion first (cheap, local); only if it is irreducibly fuzzy _and_ you observe the rush, hide the post-completion steps by splitting (the sequence cut).
+- **Duplication** — the same meaning in more than one place. Costs maintenance and tokens, and inflates a meaning's prominence on the ladder past its real rank.
+- **Sediment** — stale layers that settle because adding feels safe and removing feels risky. The default fate of any skill without a pruning discipline.
+- **Sprawl** — a skill simply too long, even when every line is live and unique. Hurts readability and maintainability and wastes tokens. The cure is the ladder: disclose **reference** behind pointers, and split by **branch** or sequence so each path carries only what it needs.
+- **No-op** — a line the model already obeys by default, so you pay load to say nothing. The test: does it change behaviour versus the default? A weak leading word (_be thorough_ when the agent is already thorough-ish) is a no-op; the fix is a stronger word (_relentless_), not a different technique.
+- **Negation** — steering by prohibition backfires: _don't think of an elephant_ names the elephant and makes it more available, not less. Prompt the **positive** — state the target behaviour so the banned one is never spoken; keep a prohibition only as a hard guardrail you can't phrase positively, and even then pair it with what to do instead.

package/.agents/skills/writing-shape/SKILL.md

@@ -0,0 +1,79 @@
+---
+name: writing-shape
+description: Writing, exploit — shape raw material into an article, paragraph by paragraph.
+disable-model-invocation: true
+---
+
+<what-to-do>
+
+The user has passed (or will pass) a markdown file of raw material. Treat it as the input pile — anything from a tidy list of fragments to a wall of unstructured prose to a transcript. The format does not matter. Read it end-to-end before doing anything else.
+
+Then run a shaping session that produces a separate article document. This is **exploit**: the exploring is done, the pile is fixed — commit to a structure and mine the pile to fill it. Do not edit the raw material file — it is read-only to this skill.
+
+If the user did not say where to save the article, ask once and remember the path.
+
+</what-to-do>
+
+<supporting-info>
+
+## The loop
+
+1. **Read the pile.** Read the input file in full. Form a sense of what's in it.
+2. **Establish the prerequisites.** Settle with the user what the reader knows walking in — the concepts that are **grounded** from the start. Everything else must be grounded by a block before a later block can lean on it. See [Grounding](#grounding).
+3. **Draft 2–3 candidate openings.** Each opening should imply a different thesis or angle for the article. Show all of them. Force the user to pick or compose a hybrid. The chosen opening defines what the rest of the article must do.
+4. **Grow paragraph by paragraph.** After the opening lands, ask "given this opening, what does the reader need to hear next?" Pull material from the pile to answer. The next block may only lean on grounded concepts, and grounds new ones as it lands. Argue about the form the next block takes — a paragraph, a list, a table, a callout, a quote, a code block. Each format choice should be deliberate and defensible.
+5. **Append to the article file as you go.** Don't batch. Write each agreed paragraph or block immediately so the user can see the article taking shape.
+6. **Loop step 4 until the article is done.** The user decides when it's done.
+
+## Grounding
+
+Every **concept** has to be **grounded** before a block can lean on it: the reader either walked in knowing it or met it in an earlier block. A block that reaches for an ungrounded concept loses the reader. The unit is the concept, not the word for it — a block can lean on an idea the reader lacks even with no jargon in sight. Where a concept has a name — a **term** — grounding it means landing the idea and the term together.
+
+A concept gets grounded one of two ways:
+
+- **Prerequisite** — grounded before the opening. The reader brings it. Fixed at the start.
+- **Introduced** — a block establishes it, and from then on it's grounded for the rest of the article.
+
+Keep a running list of what's grounded. When you ask "what does the reader need to hear next?", an ungrounded concept the next move needs is itself the answer: ground it first — here or in an earlier block — or you can't make the move. This is the gap-naming of [Pulling from the pile](#pulling-from-the-pile) one level up: there the pile is missing material; here the article is missing a foundation.
+
+The lever is what you make a prerequisite versus what you ground inside the article. Demand too much up front and you shut readers out; ground too much inside and the opening drowns in definitions. Settle it with the user when you establish prerequisites.
+
+## Conversational feel
+
+This is a grilling session inverted. In ideation, the question was "what are you actually noticing?" Here it's "what is this article actually arguing, and in what order does the reader need to hear it?" Push back. Refuse to let weak transitions slide. If a paragraph doesn't earn its place, cut it.
+
+Specific moves to keep using:
+
+- "What does this paragraph do for the reader that the previous one didn't?"
+- "If I cut this, what breaks?"
+- "Is this prose, or should it be a list? Why prose?"
+- "This sentence is doing two jobs — split it or pick one."
+- "The opening promised X. We've drifted to Y. Either re-thread it or change the opening."
+
+## Pulling from the pile
+
+Treat the raw material as a quarry, not a script. Pull a fragment, rework it to fit the surrounding paragraph, and place it. A fragment may be split across multiple paragraphs, merged with another, or paraphrased. The pile's job is to be mined; the article's job is to read as one voice.
+
+If the pile lacks something the article needs, name the gap explicitly: "We need an example here and the pile doesn't have one — give me one now or we cut this section."
+
+## Format arguments to actually have
+
+When choosing how to render a block, weigh these tradeoffs out loud with the user, not silently:
+
+- **Prose vs. list.** Prose carries argument; lists carry parallel items. If items aren't truly parallel, prose is better. If they are, a list is faster to scan.
+- **Inline vs. callout.** Tips, warnings, and asides go in callouts (`> [!TIP]`, `> [!NOTE]`) — but only if they'd genuinely derail the main argument inline. Otherwise leave them inline.
+- **Table vs. repeated structure.** If the same shape repeats 3+ times with the same fields, a table. Otherwise prose with bold leads.
+- **Quote vs. paraphrase.** Quote when the original wording is the point. Paraphrase when only the idea matters.
+- **Code block vs. inline code.** Multi-line, runnable, or illustrative → block. Single token or identifier → inline.
+
+## Writing rhythm
+
+Append to the article file as each block is agreed. Re-read the file from disk before every write — the user may have edited between turns. Never overwrite blindly. If the user wants a paragraph rewritten, edit that specific paragraph in place; leave the rest alone.
+
+## Out of scope
+
+- Mining for new fragments that aren't in the pile (handle gaps as in "Pulling from the pile").
+- Editing the raw material file.
+- Publishing, formatting for a specific platform, or adding frontmatter the user didn't ask for.
+
+</supporting-info>