Add a team of AI agents to your MkDocs site — a support agent grounded in your docs that answers with cited sources and can reach your tools via MCP, asking before it changes anything. Because MkDocs builds static HTML, embedding is a single script tag.
What your readers get
A reader who's stuck gets an instant, cited answer pulled from your own documentation, not a generic guess. Behind the widget, AIML.chat runs a Support agent grounded in your content and tool-calling agents that act through the Model Context Protocol (MCP), with human write-confirm on anything that changes data. Every answer links back to the exact page it came from, so trust stays high — and the analytics show which questions go unanswered, telling you what to document next.
Add it in three steps
- 1Create a website in AIML.chat and point the crawler at your published docs URL (or connect the GitHub repo your Markdown lives in). It indexes and embeds every page automatically.
- 2Add the embed snippet to your theme's head. With Material for MkDocs, drop it into an
overrides/main.htmlblock ({% block extrahead %}); on a custom theme, add it to the base template's<head>. - 3Rebuild and deploy. The widget appears site-wide, themed to match, and stays current as you re-crawl.
Because the widget is vanilla JS under 30KB gzipped, it won't bloat your build or fight Material's theme. As your docs change, a re-crawl — manual, scheduled, or via the GitHub Action on every push — keeps answers accurate. For other doc tools, see the VitePress and Docusaurus guides.
Do I need an MkDocs plugin?+
No. MkDocs outputs static HTML, so the widget embeds with one script tag added to your theme head — no plugin required.
Does it work with Material for MkDocs?+
Yes. Add the snippet via an overrides/main.html extra-head block and it loads site-wide, themed to match.
Is there a free plan?+
Yes — 100 messages a month and one website on the free tier. See pricing for higher limits.