What exactly does Kodadocs generate?
It creates a full VitePress help site with articles like Getting Started, Feature Guides, FAQ, and Troubleshooting, all written by AI and accompanied by annotated screenshots.

@Kodadocs
Basic information
AI-generated help docs for your web app in minutes, not weeks. KodaDocs is a Claude Code MCP server that reads your codebase, captures screenshots with a headless browser, annotates UI elements, writes documentation articles, and assembles a complete VitePress help center. Suppor
Add this server to your MCP-compatible client using the configuration below.
{
"mcpServers": {
"kodadocs": {
"command": "uvx",
"args": [
"kodadocs",
"mcp"
]
}
}
}Detect the web framework of a project at the given path. Returns the framework name (e.g., 'Next.js', 'Django', 'React'). Uses heuristic detection from package.json, requirements.txt, etc.
Capture screenshots for discovered routes using Playwright. Launches headless Chromium, authenticates if auth is provided, navigates to each route, captures screenshots, and extracts DOM elements. Returns JSON with screenshots dict and dom_elements dict.
Annotate screenshots with numbered callouts using Pillow. Takes screenshot directory and dom_elements dict (route -> element list). Each element needs: role, name, bounds {x, y, width, height}. Creates annotated copies in screenshots_dir/annotated/. Returns JSON with route -> annotated image path.
Assemble a VitePress static site from generated articles and screenshots. Creates index page, article markdown files, VitePress config, theme with brand color, and package.json. Pass theme_name to use a built-in theme preset (default, professional, minimal, playful, dark-modern, docs-classic). Pro themes require a valid license_key.
Save pipeline state to .kodadocs/run_manifest.json.
Load existing pipeline state from .kodadocs/run_manifest.json.
Deploy the generated VitePress site to a hosting provider. Supported providers: cloudflare, vercel, netlify, github-pages, kodadocs. For kodadocs provider: deploys to <site_slug>.kodadocs.com. Requires license_key (or KODADOCS_LICENSE_KEY env var) and site_slug. Pass license_key for other providers to remove the KodaDocs badge.
Discover application routes, services, and metadata via static analysis. Returns JSON with discovered_routes, route_metadata, detected_services, ui_components, and deployment_platform. Supports Next.js, SvelteKit, Nuxt, React Router, and WordPress. Pass app_url to enable Playwright crawler fallback when static analysis finds few routes.
Analyze codebase structure using tree-sitter parsing. Extracts code chunks (functions, classes), error patterns, and data models (Prisma, Drizzle). No AI calls — fully deterministic. Returns JSON with code_chunks, error_patterns, data_models, and file counts.
Kodadocs is an open-source MCP server for Claude Code that automatically generates a complete help center for any web application. It produces polished documentation with AI-written articles and annotated screenshots.
Install the package and Chromium browser, then ask Claude in Claude Code: “Generate help docs for my app.” The server handles the entire pipeline from discovery to deployment.
It creates a full VitePress help site with articles like Getting Started, Feature Guides, FAQ, and Troubleshooting, all written by AI and accompanied by annotated screenshots.
Run `pip install kodadocs` and then `playwright install chromium`. No other dependencies are needed.
Next.js, Nuxt, React, Vue, Angular, SvelteKit, Remix, Astro, Django, Flask, FastAPI, Rails, Laravel, Express, Hono, WordPress, and more.
Yes—the headless browser capture tool includes support for authentication and also blurs personally identifiable information (PII).
Yes, it is open-source. The MCP server itself is free; you only pay for Claude Code usage when generating docs.
Context7 Platform -- Up-to-date code documentation for LLMs and AI code editors

Exact, shared memory for AI agents: a hierarchical database where totals always add up, every write is attributed to the agent and person behind it, and any figure traces back to its source file. Agents create databases,

Sooveryn gives Claude Code a team of AI personas that remembers your project. Soov, the partner and team lead, keeps your decisions and how you work in a lasting memory, and finds them again by words and by meaning, from

Islamic source data for assistants: Quran, hadith, tafsir, duas, prayer times and qibla. Answers come from real sources with citations instead of recalled text, so verses and hadith are not hallucinated. 36,000+ hadiths

MCP connector for Snipara, the Project Brain for AI coding agents
Comments