doc2mermaid 0.1.2

Convert documents to visual knowledge maps via LLM + Mermaid

doc2mermaid

Convert documents to visual knowledge maps — powered by LLM + Mermaid.

Document Text → LLM Structure Extraction → Graph JSON → Mermaid → SVG/PNG

Install

pip install doc2mermaid

# Mermaid CLI for SVG/PNG rendering
npm install -g @mermaid-js/mermaid-cli

Quick Start

Python API

from doc2mermaid import doc_to_map

svg_path = doc_to_map(
    open("article.md").read(),
    output="map.svg",
    llm_base_url="https://api.openai.com/v1",
    llm_api_key="sk-...",
    llm_model="gpt-4o-mini",
)

CLI

export DOC2MERMAID_BASE_URL=https://api.openai.com/v1
export DOC2MERMAID_API_KEY=sk-...
export DOC2MERMAID_MODEL=gpt-4o-mini

doc2mermaid article.md -o map.svg
doc2mermaid article.md -o map.png --theme dark
cat article.txt | doc2mermaid - -o map.svg

How It Works

1. Extract: LLM reads your document and extracts a graph structure (nodes + edges)
2. Render: The graph is converted to Mermaid diagram syntax with auto-styling
3. Output: Mermaid CLI renders the final SVG or PNG

Node Types & Colors

Type	Color	Use For
problem	Red	Problems, challenges, pain points
idea	Purple	Ideas, hypotheses, concepts
method	Blue	Methods, approaches, solutions
step	Cyan	Process steps, actions
result	Green	Results, outcomes, findings
takeaway	Yellow	Key takeaways, conclusions

Configuration

LLM settings via environment variables or function arguments:

Env Var	Argument	Description
DOC2MERMAID_BASE_URL	llm_base_url	OpenAI-compatible API base URL
DOC2MERMAID_API_KEY	llm_api_key	API key
DOC2MERMAID_MODEL	llm_model	Model identifier

Works with any OpenAI-compatible API (OpenAI, Claude, DeepSeek, local models, etc.).

License

MIT