Skill Progressive Disclosure Design

Skill Progressive Disclosure Design

Each section that recommends a direction includes explicit pros and cons. The decisions in this skill are trade-offs, not rules. The model using this skill should reason from the trade-offs to the user's specific situation rather than apply rules blindly.

Triggering vs. disclosure: separate these first

Two problems get conflated and need separating before any splitting decision.

Triggering is whether Claude invokes the skill at all. Driven entirely by the YAML description. File splitting does not affect triggering. If the question is "my skill doesn't trigger reliably", do not split files, fix the description (use run_loop.py from the skill-creator skill).

Progressive disclosure is what loads after the skill activates. SKILL.md body always loads. references/* only loads when SKILL.md tells the model to read a specific file. scripts/* executes without loading into context at all. This is where context protection happens.

If the user is asking about splitting because of triggering issues, surface the confusion first and redirect.

Default: do not split

A monolithic SKILL.md beats a split one until proven otherwise.

Split only when at least one is true:

• SKILL.md exceeds ~400 lines and content has natural branches.
• Empirical evidence (eval transcripts) shows the model wasting context on irrelevant sections.
• Specific content is large and only needed in narrow conditions.

Pros of staying monolithic:

• Single context load, no router prose to maintain.
• No tool-call overhead from reading references.
• No risk of the model loading the wrong reference or skipping a needed one.
• Easier to maintain: one file, one source of truth.
• Better for highly interconnected content where context is global.
• Easier for human reviewers to read end-to-end.

Cons of staying monolithic:

• Every invocation pays the full token cost, even when only 10% of the content is relevant.
• Does not scale past ~500 lines without degrading the model's ability to find what matters.
• No mechanism to gate rare or niche content.
• All content must justify its always-loaded status.
• Maintenance gets harder as the file grows.

Three split axes that work

1. Variant branch

User intent selects exactly one path. SKILL.md holds the decision logic and shared workflow. Each references/<variant>.md holds path-specific detail.

my-skill/
├── SKILL.md               # decision tree + shared steps
└── references/
    ├── variant-a.md
    ├── variant-b.md
    └── variant-c.md

Examples of clean variants: cloud provider, database engine, framework choice, output format, language.

Pros:

• Each invocation loads only the matching variant; large savings when variants are big.
• Variants evolve independently, simplifying maintenance.
• Adding a new variant does not bloat existing content.
• Mental model is easy: select one path based on input.
• Maps cleanly to user intent that already mentions the variant.

Cons:

• Requires routing logic in SKILL.md, eating back some of the line savings.
• Cross-cutting changes touch every variant file, multiplying effort.
• Risk of treatments diverging across variants over time.
• If user intent is ambiguous, the model may load multiple variants and lose the savings.
• If variants share more than ~60% of their content, the abstraction breaks down.

2. Workflow vs. reference data

SKILL.md holds the procedure (verbs, sequence, decisions). references/ holds lookup material queried by key.

Good reference content: schemas, error code tables, API surface listings, example galleries, configuration option matrices, design tokens.

Pros:

• Highest leverage of all splits: lookups are narrow, the model reads one entry.
• Natural conceptual boundary (procedure vs. data).
• Reference can grow large without affecting per-invocation cost.
• Adding new reference entries does not touch the workflow.
• Reference data can often be machine-generated and regenerated.

Cons:

• The model must know what to look up before reading. Pointer must encode lookup keys explicitly.
• Fails when the workflow needs to weave reference data inline rather than at discrete points.
• Splits content that is conceptually unified, harder for human readers.
• The model may miss broader context that lives only in the reference.
• Lookup data that is small (under ~50 lines total) is rarely worth splitting.

3. Depth tier (common path vs. edge cases)

SKILL.md covers the 80% case. references/edge-cases.md covers the rest.

The pointer must read like:

If you see X, Y, or Z, stop and read references/edge-cases.md before continuing.

Pros:

• Common path stays minimal, fast, cheap.
• Edge cases can be exhaustive without polluting every invocation.
• Easy to extend edge-case coverage without touching the common path.
• Mirrors how experts work: defaults first, exceptions on demand.

Cons:

• The load condition must be sharp and observable from user input. Most edge cases do not satisfy this.
• Vague conditions cause either always-loading (waste) or never-loading (dead weight).
• Edge cases get less testing because evals naturally cluster on common queries.
• The model may follow the common path past a point where it should have escalated.
• The 80/20 estimate is often wrong; what looked like an edge case turns out to be common.

Splits that do not work

For each anti-pattern, "why it appears attractive" shows what makes designers reach for it; "why it fails" shows what goes wrong in practice.

Topic-based splits where invocations do not cluster by topic

A testing skill split into unit.md, integration.md, mocks.md is a typical example.

Why it appears attractive:

• Conceptually clean, mirrors how a human would organize documentation.
• Easy to navigate as a maintainer.
• Plausibly reduces context per invocation.

Why it fails:

• Real tasks span 2-3 topics, forcing multiple loads per invocation.
• Cross-topic concerns get duplicated or fragmented.
• The savings are theoretical, not empirical.

Splitting to hit a line target without a real branching condition

Why it appears attractive:

• A heuristic ("keep SKILL.md under 400 lines") feels like a clean rule to satisfy.
• Splitting feels like progress.

Why it fails:

• Without a branching condition, references load in parallel or always, providing no savings.
• Adds router prose to SKILL.md, often making the total content longer.

Rare-but-critical content in references/

Why it appears attractive:

• The content is large or specialized.
• Moving it out of SKILL.md feels like good hygiene.

Why it fails:

• References are optional by design; the model may skip them.
• If the content is critical, it must be loaded reliably, which means SKILL.md.
• "Rare" and "critical" together is usually a sign the skill is doing two jobs and should be two skills.

Cosmetic splits (Examples, Notes, Tips files)

Why it appears attractive:

• Reduces visual clutter in SKILL.md.
• Feels like good organization.

Why it fails:

• No load condition: either always loaded (wasted tool call) or never loaded (dead content).
• Implies an importance hierarchy that does not exist at runtime.
• Frequently hides content from the model that needs it.

Pointer hygiene

When SKILL.md points at a reference, the pointer is the entire load contract. Rules:

• Name the user-visible signal that triggers the load. "If the user mentions snapshot tests" not "for testing concerns".
• One sentence per pointer. Do not summarize the reference content in SKILL.md.
• Encode the load condition in the filename. go126-simd.md not advanced.md.
• Top-of-file table of contents for any reference over 300 lines.
• If two references are co-loaded in most runs, merge them.

Pros of strict pointer hygiene:

• Wrong-load rate drops sharply.