Henry34 downloadsResurface contradictions, gaps, and forgotten connections from your notes. Keep your knowledge in a database you own, where any LLM can build on it. Extraction is automatic, decisions are yours.
Your forgotten notes start talking back. OntoFlow reads your vault and returns what you could not see: claims that contradict each other, concepts every note links to but none explains, and the same idea written twice in different places. The machine organizes. You decide.
Every finding is grounded in your own notes. One contradiction, one gap, one unexpected connection: from the very first indexing screen, everything shown comes from notes you wrote. No invented examples.
The engine behind all of this is a knowledge graph with a closed verb registry: notes connect by meaning (supports, contradicts, is part of, causes, ...), and every machine suggestion waits for your confirmation in the review queue. The graph is the engine, not the billboard: what you see are findings, not pictures.
Notes outside Obsidian (plain Markdown folders, Logseq, Bear, OneNote, Evernote, Notion, Google Drive, Dropbox, GitLab, GitHub, Jira and Confluence) join through a single Connect button, with no manual token setup (only Notion takes an internal integration token).
Install from the Obsidian community plugin directory, or manually: download
main.js and manifest.json from a release into
<your vault>/.obsidian/plugins/ontoflow/, then refresh and enable the plugin
under Settings → Community plugins.
Minimum Obsidian version: 1.8.7. Works on desktop and mobile. Collecting from folders outside the vault (plain Markdown, Logseq, synced folders) and from Bear requires file-system access and is desktop-only; on mobile those items are explicitly marked unavailable rather than silently empty.
By default, nothing leaves your device. Installing and enabling the plugin sends nothing anywhere. Until you enter a hub address and token and complete the sensitive-data consent, your notes stay on this device. The plugin contains no telemetry, no analytics, no ads, and no path that downloads and runs remote code (even the SQLite engine's wasm is embedded in the bundle).
After you connect, the plugin talks only to:
| Where | What | When |
|---|---|---|
| Moayo hub (the address you enter in settings) | Note content to extract, structured results | During indexing and collection. The hub keeps zero content (zero retention) |
| Your BYO endpoint (the address you enter) | Note content to extract | Only if you configure BYO. In this mode note content never leaves your machine |
| Your own PostgreSQL (via the hub) | Your graph database | Only if you choose pg as your database |
data.json) and flow only as request headers. They
never appear in note content, logs, or error messages.Apache-2.0. See LICENSE.
This repository is the public source of the OntoFlow Obsidian plugin: the plugin itself plus the core modules its build needs. It is synced from the OntoFlow monorepo. Issues and pull requests are welcome here.