Documentation skills — Mermaid, PlantUML, ProcessOn, API docs, full-stack-doc, MarkItDown (awesome / cli / ocr)
English | 简体中文
Introduction · Install · Skills · Usage Guide · Supported Agents · Ecosystem
Documentation Skills is a curated collection of Agent Skills for AI coding agents, published in the Full Stack Skills ecosystem.
This package includes 11 skills. Each skill is a self-contained SKILL.md file that AI agents load on-demand.
npx skills add full-stack-skills/document-skillsOr install specific skills:
npx skills add full-stack-skills/document-skills --skill <skill-name>| Skill | Description |
|---|---|
api-doc-generator |
Generate traceable API documentation from OpenAPI, routes, source, and tests |
doc-coauthoring |
Gather context, draft collaboratively, and run independent reader tests |
full-stack-doc |
Build an example-rich documentation system from lifecycle, README, Rust-profile, and complete architecture template families |
markitdown-awesome |
Overview & navigation for Microsoft MarkItDown — capabilities, supported formats, three-path decision tree (awesome / cli / ocr), and side-by-side comparison with textract / unstructured / pandoc / docling / Azure DI / Azure CU |
markitdown-cli |
Complete CLI guide for markitdown — every flag (-o/-x/-m/-c/-d/-e/-p/--use-cu/--cu-endpoint/--cu-analyzer/--cu-file-types/--list-plugins/--keep-data-uris), I/O matrix, shell-toolchain integration, Azure backends, troubleshooting |
markitdown-ocr |
Drive the markitdown-ocr plugin to run LLM Vision OCR (OpenAI-compatible, including Azure OpenAI / Gemini / local vLLM / Ollama) on PDF / DOCX / PPTX / XLSX embedded images and scanned pages — install, configuration, Python API, scanned-PDF full-page fallback, per-format behavior, custom prompts, cost & failure handling |
mermaid |
Create Mermaid diagrams that render directly in Markdown and README files |
plantuml |
Create precise UML, C4, and enterprise architecture source |
processon-diagram-generator |
Generate editable DSL and rendered images through the ProcessOn API |
processon-mindmap |
Organize a hierarchical mind map ready for ProcessOn |
technical-blog-doc |
Write reproducible technical tutorials with environment, commands, and evidence |
full-stack-doc includes the original detailed root, version, module, and delivery templates plus Java, Rust, plugin, skill-ecosystem, and complete-reference README templates. The Rust common template composes with profiles for file formats, upstream parity, toolbox workspaces, authentication frameworks, design-only repositories, and multilingual layouts. Its complete architecture master composes with runtime, plugin, edge, event, AI/RAG, and observability/control-plane profiles. API discovery loads focused guidance for OpenAPI, Spring, FastAPI, NestJS, Express, or Gin.
After installation, name the skill explicitly and describe the expected artifact, source material, output location, and verification requirements. Client syntax varies: some clients expose /skill-name, Codex-compatible prompts may use $skill-name, and clients with automatic discovery also accept the plain skill name.
These three prompts are equivalent in intent:
/full-stack-doc Create a detailed architecture document from the current repository.
$full-stack-doc Create a detailed architecture document from the current repository.
Use the full-stack-doc skill to create a detailed architecture document from the current repository.
The examples below use /skill-name; replace the prefix when your client uses another form.
| Skill | Use it for | Example prompt |
|---|---|---|
api-doc-generator |
API inventories and evidence-backed endpoint references | Use /api-doc-generator to generate API documentation from the current Spring controllers, DTOs, and tests, with a source for every endpoint. |
doc-coauthoring |
Collaborative proposals, RFCs, specifications, and reader tests | Use /doc-coauthoring to help me draft a cache-refactoring RFC and run an independent reader review. |
full-stack-doc |
Product documentation suites, PRDs, architecture, UI, versions, delivery docs, and project READMEs | Use /full-stack-doc to audit the existing product-docs tree and complete the missing product-level and version-level documents. |
markitdown-awesome |
MarkItDown capability discovery and tool selection | Use /markitdown-awesome to compare MarkItDown, Pandoc, and Docling for our PDF-to-Markdown workflow. |
markitdown-cli |
Exact markitdown commands, flags, pipelines, and batch conversion |
Use /markitdown-cli to provide commands that batch-convert DOCX and PPTX files under docs/. |
markitdown-ocr |
OCR for scanned pages and images embedded in PDF or Office files | Use /markitdown-ocr to convert this scanned PDF to Markdown and document model configuration, cost, and fallback behavior. |
mermaid |
Markdown-native flows, sequences, states, journeys, timelines, and lightweight architecture | Use /mermaid to turn the order-refund flow into a sequence diagram that renders in README.md. |
plantuml |
Precise UML, C4, deployment, and enterprise architecture source | Use /plantuml to generate C4 container and deployment diagrams for the authentication system. |
processon-diagram-generator |
Hosted, editable ProcessOn diagrams and optional rendering | Use /processon-diagram-generator to create an editable login swimlane diagram and return its DSL and rendered output. |
processon-mindmap |
ProcessOn-ready hierarchical mind-map outlines | Use /processon-mindmap to organize these meeting notes into a ProcessOn-ready mind map. |
technical-blog-doc |
Reproducible tutorials, integration guides, deployment articles, and verification records | Use /technical-blog-doc to write a reproducible deployment tutorial from the current repository and include a verification record. |
processon-diagram-generator requires PROCESSON_API_KEY for remote generation. The other examples can begin with local source or material supplied in the conversation; access to external services still depends on the active client and environment.
Use /full-stack-doc to create a detailed architecture document.
Use /full-stack-doc to create detailed English and Chinese README documents.
Use /full-stack-doc to create a detailed technical deep-dive document.
Use /full-stack-doc to initialize the 10 product-level documents and the 7 V1 documents.
Use /full-stack-doc to create a V2 PRD, version architecture, and feature-menu plan from evidence in the current repository.
Use /full-stack-doc to audit the current product-docs structure, naming, cross-references, and content gaps, separating facts, inferences, and open questions.
For more reliable output, make the request concrete:
Use /full-stack-doc with {repository or source material} for {target readers} to create {document type} in {output directory}. Inspect existing documents and source first, preserve confirmed conventions, mark assumptions and open questions, and run the applicable validation checks before delivery.
python3 scripts/validate_repository.py
python3 skills/full-stack-doc/scripts/validate_templates.py
python3 -m unittest discover -s skills/full-stack-doc/tests -p 'test_*.py'
python3 -m unittest discover -s skills/api-doc-generator/tests -p 'test_*.py'
python3 -m unittest discover -s skills/processon-diagram-generator/tests -p 'test_*.py'
python3 scripts/run_artifact_evals.pyTrigger positive/negative cases and artifact regression cases live under evals/.
Works with Claude Code, Codex, Cursor, OpenCode, Gemini CLI, GitHub Copilot, Windsurf, and 70+ others.
Option 1: npx skills CLI (Recommended)
npx skills add full-stack-skills/document-skillsOption 2: Manual Installation
git clone https://github.com/full-stack-skills/document-skills.git
cp -r document-skills/skills/* .claude/skills/For more details, see the Claude Code Skills Guide and Agent Skills Spec.
| Resource | Link |
|---|---|
| Full Stack Skills | github.com/full-stack-skills |
| All Skill Groups | github.com/full-stack-skills |
| Agent Skills Spec | agentskills.io |
| Skills CLI | github.com/vercel-labs/skills |
Apache 2.0 — see LICENSE.