Programmatic Doc Generation
April 27, 2026 · View on GitHub
A Claude Code plugin for building programmatic document generation pipelines.
Two complementary tracks:
- Local Typst typesetting for small-batch script-driven runs — author a parameterised
.typtemplate once, render N PDFs from N JSON data files. - Automation-pipeline integration — wire document generation into n8n / serverless / API workflows using one of seven supported rendering engines, with reference docs fetched fresh and stored alongside the user's saved engine preference.
The plugin can ingest existing billing/business document samples (PDF, image, Word), extract a JSON schema, and convert them into Typst templates ready for batch rendering. Brand identity (palette, fonts, logo, business address) is captured once and reused across every template.
Skills
new-typst-template— scaffold a parameterised Typst template for batch rendering.batch-render-typst— render N PDFs from a directory of JSON data files against a Typst template.ingest-document-sample— turn an existing PDF/image/DOCX sample into a Typst template plus JSON schema.schema-from-samples— infer a unified JSON schema from multiple samples of the same document type.define-brand-profile— capture brand palette, logo, business identity into a reusable profile.extract-brand-from-site— auto-derive a brand profile from a website URL.recommend-rendering-service— pick the right rendering engine for the user's stack and produce a worked integration example.scaffold-n8n-doc-pipeline— generate an importable n8n workflow JSON for webhook → render → store.
Slash commands
/doc-gen-onboard— interactive setup: pick a default rendering engine, fetch its reference docs, persist the choice.
Supported rendering engines (out of the box)
| Engine | Tier | Best for |
|---|---|---|
| Typst CLI | Free / self-host | Local batch, dev-driven pipelines |
| Carbone | Generous free + OSS core | DOCX/ODT templates with JSON binding |
| PDFMonkey | Generous free tier | HTML/Liquid, native n8n integration |
| APITemplate.io | Small free + cheap paid | HTML + JSON, also images |
| DocRaptor | Paid (free dev mode) | Print-grade HTML/CSS via PrinceXML |
| Docmosis | Enterprise | High-volume regulated billing |
| Adobe Document Generation API | Enterprise | Word templates + Acrobat Sign |
Templafy (Office template governance) and PandaDoc (sales contracts + e-sign) are surfaced contextually rather than as defaults.
Data storage
The plugin follows the standard data-storage rules for Claude Code plugins:
- User preferences and reference docs live under
${CLAUDE_USER_DATA:-${XDG_DATA_HOME:-$HOME/.local/share}/claude-plugins}/programmatic-doc-generation/. - User-owned outputs (templates, rendered PDFs, n8n workflows) live wherever the user chooses — typically under
~/Documents/or~/repos/. The plugin stores only path pointers. - The plugin install directory is never written to (it's clobbered on updates).
Installation
claude plugins marketplace add danielrosehill/Claude-Code-Plugins
claude plugins install programmatic-doc-generation@danielrosehill
After installation, run /doc-gen-onboard to pick a default rendering engine.
License
MIT.