Read and write pages— Have your assistant fetch any Markdown page's content or create/update pages directly through the wiki's MCP server, with changes reflected in the live index.

Explore the knowledge graph— Ask for related pages, backlinks, or a focused view of how a page connects to others via links, folders, and shared tags.

Manage files and structure— Instruct your assistant to rename, move, or copy pages and folders across Spaces, with git history tracking every change as a proper rename.

Invoke external MCP tools— Let your assistant call out to registered MCP servers (search engines, ticketing systems) during a response, with per-server permissions you've defined.

Save tool results as pages— Use the Tool Explorer to make deterministic calls to any MCP server's tools and save the output directly as a new wiki page.

Everything your team needs to stay organised

Twenty tightly integrated features with AI woven throughout — assistants, agents, MCP, and a knowledge graph. Click any card to read the full feature guide.

Write in Markdown with a full-featured editor. Classic and inline editing modes, keyboard shortcuts, find & replace, content transclusion, sequence diagrams and flowcharts written as text, and print-ready pages.

Create and edit diagrams directly inside the wiki using the embedded draw.io editor. Diagrams render as SVG previews in pages and can be embedded inline with a simple tag.

Spreadsheet-style data tables with typed columns (text, date, choice, auto-increment), named views, filters, column ordering, and CSV / XML export.

Threaded chat as standalone files in the wiki tree — real-time polling, emoji picker, user mentions with autocomplete, message grouping, and optional topic banners.

A chat panel attached directly to any Markdown page. AI assistants receive the full page content as automatic context — and if they update the page, the viewer refreshes live with a toast notification.

Full-text search across all pages with result previews, timestamps, and author attribution. Tag any page and browse the interactive tag cloud to filter by topic.

Save a query as a reusable.searchpage — mix full-text withtag:,updated:, and MCP source filters. It remembers its last result so you reopen an answer, not a blank box.

Divide your wiki into isolated workspaces — one per team, project, or department. Switch instantly from the sidebar, with per-user access control for each space.

Content is plain files, so edit it however you like — rsync, git pull, a desktop editor over a share. The wiki notices: the index, search and file tree follow, and the page you're reading reloads itself. No cron job, no watcher to install.

Attach any file to any page by drag-and-drop or file picker. Attachments are stored per-page, access-controlled, and available for direct download.

Optional OIDC / Auth0 single sign-on with three roles (admin, editor, viewer) plus anonymous reader access for public browsing. Per-space restrictions, session timeout, and email notifications for access requests.

Create folders and nested hierarchies. Rename, copy, and move any page or folder — including across Spaces. All operations keep the page index in sync automatically.

Add AI bots backed by Claude, GPT, or any OpenAI-compatible model. Trigger them with @mentions in any chat thread, or open a Page Chat panel on any Markdown page for AI with automatic page context. Bots can read and write wiki pages as tools during a response.

Automatic commits on every save, with renames and moves tracked as proper git renames so history follows files across reorganisations. Manual snapshots for chats and lists. One-click restore for any version.

One container — nginx, PHP and cron — with every page, setting and log kept on the host. Configure it with a single environment file, back it up with tar, and put it behind HTTPS in minutes.

Full UI translation in English, Danish, Swedish, Spanish, French, German, Simplified Chinese, and Hindi. Each user picks their language from the sidebar — stored in the browser, no account required.

Schedule AI agents to run automatically — daily, on specific weekdays, or monthly. Each job uses an agentic loop to read and write wiki pages, with failure alerts sent by email.

Start new pages and diagrams from structured templates. Ten bundled templates cover meeting notes, decision records, how-to guides, API references, and more. Add your own per-space templates at any time.

Send any page to colleagues with one click. Choose everyone or specific recipients with type-ahead search. The e-mail includes a deep link that opens the exact page in the correct Space.

Let visitors browse and read the wiki without logging in. Authentication still protects editing; reader access can be enabled globally or restricted per space — ideal for public-facing knowledge bases.

Connect any MCP client — Claude Desktop, Claude Code, your own agent — straight to the wiki's pages. AI Users can likewise call out to external MCP tool servers as part of their replies.

Say "Remember …" and your AI Users file the fact away in a dedicated Memory space, then recall it later — a persistent, shared brain built entirely from wiki pages. No database, no extra service.

Explore your wiki as an interactive map. Pages connect by explicit links, folder structure, and shared tags — with backlinks, related-page discovery, and a per-page focus view. The same relationships give AI a structured view of how everything fits together.

Opt in to a once-a-day digest email summarising the pages created or updated in the last 24 hours across the spaces you can access — grouped by space, your own edits excluded, capped at the 20 most recent changes.

Read your wiki on a phone. A responsive, view-focused layout turns the sidebar into a slide-in menu and widens content for small screens — with a one-tap toggle to switch between automatic, desktop, and mobile.

Astucia LLM Wiki isn't just AI-enabled — it's built so AI assistants work directly inside your team's real knowledge, with the tools and boundaries you define, instead of a generic chatbot bolted on the side.

Assistance in the Context of Your Documents

Automated document processing and extraction