Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Comprehensive documentation specialist covering API documentation, technical writing, design documentation, migration guides, and changelog generation. Use when creating OpenAPI/Swagger specs, generating SDKs, writing user guides, creating README files, documenting architecture, writing design specs, creating ADRs, writing migration guides, or generating changelogs from git commits. Handles versioning, examples, developer experience, and user-facing documentation.
.claude/skills/aiskillstore-documentation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | 101% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 117% | 0% |
| case-08 | ✗→✓ | ▲ Improved | 50% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 39% | 0% |
Ensures all documentation follows project standards and prevents redundant files.
This skill provides tools and workflows for efficient documentation management:
Use this skill when:
docs/
├── README.md # Master index (DO NOT MODIFY)
├── guides/ # User and developer guides
├── architecture/ # Architecture docs
└── screens/ # Screen-specific docs
└── {screen-name}/
├── README.md # Overview (required)
├── features.md # Features (optional)
├── technical.md # Technical (required)
└── flows.md # Flows (optional)Use scripts for efficient search (80-98% context reduction):
bash# Search without reading full files ./scripts/doc-search.sh "topic-keyword" # Get file metadata (0 lines read) ./scripts/doc-metadata.sh docs/path/to/file.md # List sections to navigate (structure only) ./scripts/doc-list-sections.sh docs/file.md
Complete script documentation: See scripts/README.md
If similar documentation exists:
references/workflows/update-existing.mddoc-get-section.sh to read only relevant sectiondoc-update-section.shIf creating new documentation:
references/workflows/create-screen.mdassets/templates/screen-readme.md or assets/templates/screen-technical.mdIf file is too large (>450 lines):
references/workflows/split-large-file.mdIf duplicates found:
references/workflows/consolidate-duplicates.mdLoad reference file: references/rules.md
Check:
Complete rules: See references/rules.md
scripts/)Efficient section-level interaction tools:
Benefits: 80-98% reduction in context usage vs reading entire files
Complete documentation: scripts/README.md
references/)Load as needed for detailed information:
create-screen.md - Creating new screen documentationupdate-existing.md - Updating existing documentationconsolidate-duplicates.md - Consolidating duplicate docssplit-large-file.md - Splitting files >500 linesadd-examples.md - Adding examples to docsreview-quarterly.md - Quarterly documentation reviewassets/)Templates used in documentation output:
screen-readme.md - For docs/screens/{name}/README.mdscreen-technical.md - For docs/screens/{name}/technical.mdscreen-features.md - For docs/screens/{name}/features.mdscreen-flows.md - For docs/screens/{name}/flows.mdarchitecture.md - For docs/architecture/.mdguide.md - For docs/guides/.mdThis skill uses filesystem-based progressive disclosure:
Level 1: Metadata (always in context)
Level 2: SKILL.md (when skill triggers)
Level 3: Bundled resources (loaded as needed)
Example: Creating screen docs
./scripts/doc-search.sh "ScreenName"references/workflows/create-screen.mddocs/screens/{screen-name}assets/templates/screen-readme.mdassets/templates/screen-technical.mdreferences/rules.md./scripts/doc-metadata.sh docs/file.md./scripts/doc-list-sections.sh docs/file.md./scripts/doc-get-section.sh docs/file.md "Section"./scripts/doc-update-section.sh docs/file.md "Section" new-content.md./scripts/doc-find-duplicates.sh docs/references/workflows/consolidate-duplicates.mdBefore creating documentation:
Good documentation is:
references/structure.md)| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,814 | 29,066 | +933% | 1 | 1 | 0% | 265 | 2,099 | +692% | 0 | 0 | — |
case-02 | fail→fail | 17,672 | 6,712 | -62% | 1 | 1 | 0% | 2,551 | 2,300 | -10% | 0 | 0 | — |
case-03 | fail→fail | 8,387 | 4,875 | -42% | 1 | 1 | 0% | 1,308 | 2,048 | +57% | 0 | 0 | — |
case-04 | fail→fail | 7,698 | 5,314 | -31% | 1 | 1 | 0% | 1,270 | 2,874 | +126% | 0 | 0 | — |
case-05 | fail→fail | 12,144 | 12,852 | +6% | 1 | 1 | 0% | 1,977 | 3,934 | +99% | 0 | 0 | — |
case-06 | fail→pass | 9,301 | 7,009 | -25% | 1 | 1 | 0% | 1,405 | 2,822 | +101% | 0 | 0 | — |
case-07 | fail→pass | 7,711 | 3,216 | -58% | 1 | 1 | 0% | 1,082 | 2,349 | +117% | 0 | 0 | — |
case-08 | fail→pass | 9,517 | 2,544 | -73% | 1 | 1 | 0% | 1,503 | 2,251 | +50% | 0 | 0 | — |
case-21 | pass→fail | 14,009 | 2,452 | -82% | 1 | 1 | 0% | 2,951 | 2,144 | -27% | 0 | 0 | — |
case-09 | fail→pass | 8,911 | 2,659 | -70% | 1 | 1 | 0% | 1,525 | 2,257 | +48% | 0 | 0 | — |
case-10 | fail→pass | 8,443 | 2,585 | -69% | 1 | 1 | 0% | 1,579 | 2,198 | +39% | 0 | 0 | — |
case-11 | fail→pass | 15,361 | 4,829 | -69% | 1 | 1 | 0% | 2,188 | 2,623 | +20% | 0 | 0 | — |
case-12 | fail→pass | 17,205 | 2,810 | -84% | 1 | 1 | 0% | 2,182 | 2,165 | -1% | 0 | 0 | — |
case-13 | fail→pass | 4,780 | 2,376 | -50% | 1 | 1 | 0% | 733 | 2,141 | +192% | 0 | 0 | — |
case-14 | fail→pass | 7,635 | 1,533 | -80% | 1 | 1 | 0% | 1,124 | 2,065 | +84% | 0 | 0 | — |
case-15 | fail→pass | 13,220 | 2,973 | -78% | 1 | 1 | 0% | 1,803 | 2,258 | +25% | 0 | 0 | — |
case-16 | fail→pass | 8,883 | 3,344 | -62% | 1 | 1 | 0% | 1,347 | 2,302 | +71% | 0 | 0 | — |
case-17 | fail→pass | 6,525 | 1,640 | -75% | 1 | 1 | 0% | 1,026 | 2,109 | +106% | 0 | 0 | — |
case-18 | fail→pass | 8,966 | 2,078 | -77% | 1 | 1 | 0% | 1,451 | 2,203 | +52% | 0 | 0 | — |
case-19 | fail→pass | 8,118 | 1,719 | -79% | 1 | 1 | 0% | 1,215 | 2,128 | +75% | 0 | 0 | — |
case-20 | pass→pass | 10,482 | 24,331 | +132% | 1 | 1 | 0% | 2,069 | 6,676 | +223% | 0 | 0 | — |
case-22 | pass→pass | 16,143 | 27,500 | +70% | 1 | 1 | 0% | 3,488 | 4,940 | +42% | 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, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +59 percentage points is the difference between those two pass rates over the 19 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.