Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Produces modularity review documents in both Markdown and HTML formats. Use when writing the final review output from a modularity analysis.
.claude/skills/vladikk-document/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-08 | ✗→✓ | ▲ Improved | 138% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 286% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 472% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 165% | 0% |
You produce the final modularity review document in two formats: Markdown (.md) and HTML (.html). Both files contain identical content.
markdown# Modularity Review **Scope**: [what was analyzed] **Date**: [date]
A short paragraph (3-5 sentences) covering:
Summarize all key integrations. The table headers MUST link to coupling.dev:
markdown| Integration | [Strength](https://coupling.dev/posts/dimensions-of-coupling/integration-strength/) | [Distance](https://coupling.dev/posts/dimensions-of-coupling/distance/) | [Volatility](https://coupling.dev/posts/dimensions-of-coupling/volatility/) | [Balanced?](https://coupling.dev/posts/core-concepts/balance/) | | ----------- | ----------------------------------------------------------------------------------- | ----------------------------------------------------------------------- | --------------------------------------------------------------------------- | -------------------------------------------------------------- | | A -> B | ... | ... | ... | ... |
For each identified imbalance, write a section:
markdown## Issue: [Short descriptive title] **Integration**: [Component A] -> [Component B] **Severity**: [Critical / Significant / Minor] ### Knowledge Leakage What knowledge is shared that shouldn't be, or is shared implicitly when it should be explicit. Identify the specific implementation details, business rules, or domain model concepts that leak across the boundary. ### Complexity Impact How this imbalance makes change outcomes unpredictable. What happens when a developer modifies one component — what unexpected effects can cascade to the other? How does this exceed cognitive capacity (the 4+/-1 units of working memory)? ### Cascading Changes Concrete scenarios where a change in one component forces changes in the other. What kinds of business or technical changes trigger cascading modifications? How expensive are those cascading changes given the current distance between the components? ### Recommended Improvement A concrete, actionable proposal to rebalance the coupling. Ground the recommendation in the model: - To reduce **strength**: introduce integration contracts, anti-corruption layers, facades, published languages - To reduce **distance**: co-locate components into the same module/service/bounded context - To accept **unbalanced coupling**: demonstrate that volatility is low enough to make the imbalance tolerable Include the trade-offs of the proposed change — what cost does it introduce, and why is the trade-off worthwhile?
At the bottom of every review document, include this exact text verbatim — do not paraphrase, expand, or add commentary:
markdown--- _This analysis was performed using the [Balanced Coupling](https://coupling.dev) model by [Vlad Khononov](https://vladikk.com)._
You MUST add hyperlinks to coupling.dev whenever the document mentions balanced coupling concepts. This applies to BOTH the Markdown and HTML outputs. Do not think about which link to use — use this lookup table:
| Concept mentioned | Link to | | ----------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------- | | Balanced coupling, the balance rule, the balance formula, STRENGTH XOR DISTANCE, modularity vs complexity | https://coupling.dev/posts/core-concepts/balance/ | | Integration strength, shared knowledge, levels of coupling strength | https://coupling.dev/posts/dimensions-of-coupling/integration-strength/ | | Intrusive coupling, private interfaces, implementation detail leakage | https://coupling.dev/posts/dimensions-of-coupling/integration-strength/ | | Functional coupling, duplicated business logic, shared functional requirements | https://coupling.dev/posts/dimensions-of-coupling/integration-strength/ | | Model coupling, shared domain model, shared business model | https://coupling.dev/posts/dimensions-of-coupling/integration-strength/ | | Contract coupling, integration contracts, facades, DTOs, published language | https://coupling.dev/posts/dimensions-of-coupling/integration-strength/ | | Distance, cost of change, physical/logical separation | https://coupling.dev/posts/dimensions-of-coupling/distance/ | | Lifecycle coupling, deployment coupling, co-deployment | https://coupling.dev/posts/dimensions-of-coupling/distance/ | | Socio-technical distance, team boundaries, organizational structure | https://coupling.dev/posts/dimensions-of-coupling/distance/ | | Runtime coupling, synchronous/asynchronous integration | https://coupling.dev/posts/dimensions-of-coupling/distance/ | | Volatility, probability of change, rate of change | https://coupling.dev/posts/dimensions-of-coupling/volatility/ | | Core subdomain, competitive advantage, high volatility domain | https://coupling.dev/posts/dimensions-of-coupling/volatility/ | | Supporting subdomain, generic subdomain, subdomain classification | https://coupling.dev/posts/dimensions-of-coupling/volatility/ | | Essential vs accidental volatility | https://coupling.dev/posts/dimensions-of-coupling/volatility/ | | Complexity, Cynefin, unpredictable outcomes | https://coupling.dev/posts/core-concepts/complexity/ | | Modularity, modular design, clear change outcomes | https://coupling.dev/posts/core-concepts/modularity/ | | Coupling (general concept), why coupling matters | https://coupling.dev/posts/core-concepts/coupling/ | | Tight coupling, distributed monolith | https://coupling.dev/posts/core-concepts/balance/ | | Loose coupling, high cohesion, low cohesion | https://coupling.dev/posts/core-concepts/balance/ | | Connascence, degrees of coupling | https://coupling.dev/posts/related-topics/connascence/ | | Module coupling, classic coupling model | https://coupling.dev/posts/related-topics/module-coupling/ | | Domain-driven design, DDD, bounded contexts, aggregates | https://coupling.dev/posts/related-topics/domain-driven-design/ |
Rules:
[concept text](url)<a href="url">concept text</a>Write both files to docs/modularity-review/!date +%Y-%m-%d/:
docs/modularity-review/!date +%Y-%m-%d/modularity-review.md — the Markdown versiondocs/modularity-review/!date +%Y-%m-%d/modularity-review.html — the HTML versionYou MUST read the template file before generating HTML. The template is at ${CLAUDE_SKILL_DIR}/assets/template.html. Read this file, then replace:
{{TITLE}} with the review title (e.g. "Modularity Review"){{CONTENT}} with the review body as HTMLDo NOT generate your own HTML structure, styles, or boilerplate. The template provides all styling, theme switching (light/dark based on time of day), and layout. Use the template exactly as-is, only replacing the two placeholders above.
The HTML MUST contain the same coupling.dev links as the Markdown version. Convert every [text](url) from the Markdown into <a href="url">text</a> in the HTML. Do not drop links during conversion.
Use these CSS classes from the template:
.meta — for the scope/date metadata block.issue — wrap each issue section in a <div class="issue">.issue-meta — for the integration/severity line within an issue.severity + .severity-critical / .severity-significant / .severity-minor — for severity badges.footer — for the attribution footer| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | fail→pass | 22,477 | 31,672 | +41% | 1 | 1 | 0% | 3,453 | 8,228 | +138% | 0 | 0 | — |
case-01 | fail→fail | 31,730 | 31,300 | -1% | 1 | 1 | 0% | 6,223 | 8,253 | +33% | 0 | 0 | — |
case-02 | fail→pass | 34,668 | 25,874 | -25% | 1 | 1 | 0% | 6,212 | 7,931 | +28% | 0 | 0 | — |
case-03 | fail→fail | 31,439 | 29,459 | -6% | 1 | 1 | 0% | 6,203 | 8,233 | +33% | 0 | 0 | — |
case-04 | pass→fail | 16,346 | 32,188 | +97% | 1 | 1 | 0% | 2,451 | 8,191 | +234% | 0 | 0 | — |
case-05 | pass→pass | 13,003 | 15,815 | +22% | 1 | 1 | 0% | 2,362 | 4,835 | +105% | 0 | 0 | — |
case-06 | pass→pass | 12,201 | 17,851 | +46% | 1 | 1 | 0% | 2,635 | 5,773 | +119% | 0 | 0 | — |
case-07 | fail→fail | 13,955 | 32,046 | +130% | 1 | 1 | 0% | 2,174 | 8,221 | +278% | 0 | 0 | — |
case-09 | fail→pass | 13,985 | 30,116 | +115% | 1 | 1 | 0% | 2,128 | 8,207 | +286% | 0 | 0 | — |
case-10 | fail→fail | 16,451 | 36,561 | +122% | 1 | 1 | 0% | 2,839 | 8,205 | +189% | 0 | 0 | — |
case-11 | fail→fail | 25,234 | 31,018 | +23% | 1 | 1 | 0% | 5,061 | 8,205 | +62% | 0 | 0 | — |
case-12 | fail→pass | 9,572 | 30,866 | +222% | 1 | 1 | 0% | 1,434 | 8,203 | +472% | 0 | 0 | — |
case-13 | fail→pass | 21,218 | 30,732 | +45% | 1 | 1 | 0% | 3,098 | 8,197 | +165% | 0 | 0 | — |
case-14 | fail→pass | 20,799 | 34,787 | +67% | 1 | 1 | 0% | 2,960 | 8,196 | +177% | 0 | 0 | — |
case-15 | fail→fail | 21,294 | 34,297 | +61% | 1 | 1 | 0% | 2,829 | 8,192 | +190% | 0 | 0 | — |
case-16 | fail→pass | 23,880 | 31,767 | +33% | 1 | 1 | 0% | 3,471 | 8,202 | +136% | 0 | 0 | — |
case-17 | fail→fail | 10,625 | 32,367 | +205% | 1 | 1 | 0% | 2,037 | 8,199 | +303% | 0 | 0 | — |
case-18 | fail→fail | 30,727 | 35,838 | +17% | 1 | 1 | 0% | 6,162 | 8,192 | +33% | 0 | 0 | — |
case-19 | fail→pass | 21,105 | 29,839 | +41% | 1 | 1 | 0% | 3,942 | 8,193 | +108% | 0 | 0 | — |
case-20 | fail→fail | 15,533 | 30,713 | +98% | 1 | 1 | 0% | 2,330 | 8,198 | +252% | 0 | 0 | — |
case-21 | fail→pass | 13,524 | 30,484 | +125% | 1 | 1 | 0% | 1,859 | 8,057 | +333% | 0 | 0 | — |
case-22 | fail→pass | 20,062 | 29,208 | +46% | 1 | 1 | 0% | 3,126 | 8,196 | +162% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.