17 KiB
Glossary — Structural Predictability
The disclosed domain model for design-skill.
Every term below is a lever on Predictability.
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 because its tokens vary while its behaviour does not. 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 exposed to the model's available-skills index so the agent can discover and load it autonomously. The human can still invoke it directly, so model-invocation always includes user reach. It pays permanent context load in exchange for discoverability. A model-invoked skill whose content is all reference can provide shared reference to other discoverable skills. Pick model-invocation only when the agent must reach the skill on its own.
Avoid: ability, tool, capability
User-Invoked
A skill hidden from autonomous model discovery and reached deliberately by the human through the target harness's invocation mechanism.
It trades agent discoverability for zero context load.
Agent Skills still require a description, so use a documented harness control such as disable-model-invocation rather than deleting required metadata.
If the target harness cannot hide a skill from model discovery, report that user-only invocation is unsupported rather than inventing a mechanism.
Avoid: procedure, workflow, command
Description
Required Agent Skills metadata stating what the skill does and when it applies. For a model-invoked skill, it is the top-level context pointer kept in the model's available-skills index and the source of that skill's context load. For a user-invoked skill, it remains valid metadata but the harness keeps it out of the model's index.
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 from the context window to the 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 through its indexed description, spending both tokens and attention. A user-invoked skill escapes this load because the harness does not expose its description to the model. Context load is the brake on splitting behavior 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. It is not a cost to minimise. It is the price of human agency and 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 name other user-invoked skills and explain when the human should reach for each one. It gives the human one entry point to remember instead of many. It can only direct the human toward skills hidden from model discovery, not make those skills autonomously discoverable. It is 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 as more descriptions crowd the window and compete for attention. More user-invoked skills spend cognitive load by giving the human more 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 only when the hand-off creates a real context boundary that hides a step's post-completion steps. 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 (test-driven-development), 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, reference is secondary to them. When a skill has none, reference is the entire content. Reference can also live outside any skill entirely under 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 as a plain, non-invocable file with no description or steps. Any skill can point at it. It is the home for shared reference that need not fire on its own and can be used by skills hidden from one another's discovery surface.
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. It is 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, while 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, such as lesson, proximal zone of development, fog of war, or 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. For a model-invoked skill, the description anchors invocation. When the same word lives in prompts, documentation, and code, the agent links that shared language to the skill and fires it more reliably. Word a model-invoked skill's description with the leading words actually used to request its behavior.
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 asks whether the agent can tell done from not done. Clarity resists premature completion because a vague bound such as "understanding reached" lets the agent declare done and slip to the next step. This axis needs steps to bite because 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 strong leading word or a completion criterion that demands exhaustive work. The demand axis also applies to flat reference and drives a reference-only skill to cover every required check. 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 post-completion steps 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. It is a tug-of-war between visible post-completion steps, which pull forward, and the completion criterion's clarity, which resists that pull. A sharp, checkable bar holds while 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. Hiding works only across a real context boundary such as 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
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 an instruction still bears on what the skill does — the lens for what to keep. An instruction loses relevance either by never bearing on the task or by going stale as the behavior or world it describes changes. Shorter skills are easier to keep relevant because each instruction is cheaper to check. Distinct from no-op: relevance asks whether an instruction bears on the task, not whether it changes behavior.
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. This is the default fate of any skill without a pruning discipline. It is 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 asks whether an instruction changes behavior versus the default. An instruction 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, while No-Op is a verdict on an instruction. The two concepts 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