Overview
ondoki is a self-hosted platform for capturing, organizing, and sharing process documentation. It helps teams turn screen recordings into step-by-step guides, write rich documents, and build a searchable knowledge base — all enhanced by AI.Workflow Recording
Capture clicks, keystrokes, and screenshots from a desktop app or Chrome extension. Each step is recorded with context.
Rich Document Editor
Block-based TipTap editor with slash commands, inline AI, images, tables, code blocks, and multiple page layouts.
AI-Powered Search
Hybrid full-text + semantic vector search with reciprocal rank fusion. Find anything across documents, workflows, and knowledge sources.
Team Collaboration
Projects with role-based access, per-resource sharing, threaded comments, audit logging, and analytics.
Why ondoki?
Most process documentation tools are either too simple (screenshot-and-annotate) or too complex (enterprise knowledge management). ondoki sits in the middle:- Capture is fast — record a workflow from the desktop app and ondoki auto-generates titles, summaries, tags, and per-step descriptions using AI.
- Editing is flexible — a full block editor for when you need to write, not just capture.
- Search actually works — hybrid search combines keyword matching with semantic understanding, so you find what you need even with imprecise queries.
- AI is built in — chat with an AI that has full context on your documents and workflows. Use inline AI to write, summarize, expand, or translate directly in the editor.
- Self-hosted and open — your data stays on your infrastructure. MIT licensed.
Key Features
| Feature | Description |
|---|---|
| Workflow Recording | Desktop app and Chrome extension capture user actions with screenshots |
| Document Editor | TipTap-based block editor with slash commands, drag-and-drop, multiple layouts |
| Inline AI | Write, summarize, improve, expand, simplify, translate, explain — directly in the editor |
| AI Chat | Context-aware LLM chat with function calling (supports OpenAI, Anthropic, Ollama) |
| Hybrid Search | Full-text (tsvector) + semantic (pgvector) with RRF fusion and trigram fallback |
| Knowledge Base | Upload PDFs, DOCX, TXT, MD — content is extracted, embedded, and searchable |
| Context Links | Map URL patterns, app names, or window titles to relevant documentation |
| Video Import | Upload videos (MP4, MOV, etc.) — async pipeline extracts frames, transcribes audio, generates guides |
| MCP Server | Expose your knowledge to Claude, Cursor, or any MCP-compatible AI agent |
| Git Sync | One-way export to GitHub, GitLab, or Bitbucket as Markdown |
| Audit Log | Track all actions with filtering and CSV export |
| Analytics | Usage metrics, stale content detection, knowledge gap analysis |
| Public Sharing | Share documents and workflows via public links — no login required |
| Privacy Controls | Optional PII obfuscation via SendCloak + Presidio |
How It Works
Capture a Workflow
Use the ondoki desktop app or Chrome extension to record a process. Each click, keystroke, and navigation is captured as a step with a screenshot.
AI Processes the Recording
ondoki automatically generates a title, summary, tags, difficulty rating, and per-step descriptions using your configured LLM.
Edit and Enrich
Refine the auto-generated content in the workflow editor. Add additional steps, reorder, or merge. Write supporting documents in the rich editor.
Organize and Share
Organize content into projects and folders. Share with team members via roles, or publish public links for external access.