Obsidian Stats logoObsidian Stats

Antinomia

by Giuseppe De Martino
5
4
3
2
1
Score: 35/100
Install Code

Description

Map and work through the contradictions in your vault: tensions, tradeoffs, and conflicting goals organized in 5 layers via frontmatter, with an AI Contradiction Hunter and PDF concept extraction. Works with cloud APIs or fully local models. - This plugin has not been manually reviewed by Obsidian staff.

Reviews

No reviews yet.

Stats

stars
172
downloads
0
forks
8
days
NaN
days
NaN
days
0
total PRs
0
open PRs
0
closed PRs
0
merged PRs
0
total issues
0
open issues
0
closed issues
0
commits

Latest Version

Invalid date

Changelog

See all version on GitHub

README file from

Github

Antinomia

🎉 Now available on the Obsidian Community Store — or install from Obsidian → Settings → Community plugins → Browse → "Antinomia".

"Notes preserve. Contradictions interrogate."

Antinomia is an Obsidian plugin for Personal Tension Management (PTM) — the in-tension counterpart of Personal Knowledge Management. If PKM organizes explicit knowledge (what I know), PTM organizes where things don't fit (where something jars): contradictions, tradeoffs, anomalies, persistent doubts, weak signals, conflicts between goals. Clean ideas emerge later — as operational principles derived from resolving a tension, not before.

Obsidian Store tests License: MIT Version Obsidian Paper DOI Software DOI Substack

📓 I write about the ideas behind this work — the why, the architecture, the practice — at The Deeper Layer newsletter. Start with The Problem Nobody Is Solving and The Architecture of PTM.

✅ Status: v1.7.8 — on the Obsidian Community Store (desktop + mobile)

Antinomia is available on the Obsidian Community Store (and also installable via BRAT), on both desktop and mobile. Features are stable. Please open an issue if you find anything odd.

Antinomia is not a decision-support system. The pairs the Contradiction Hunter proposes are prompts for thinking, not truths on which to base real decisions (work, health, finance, relationships). AI models can hallucinate, oversimplify, misinterpret. Use Antinomia as a reflective practice.

The 6 layers

Every Antinomia note has a frontmatter field antinomia_type that places it in one of six layers:

Type What it is Key fields
tension A contradiction between two positions A and B status, links
substrate Raw material (quotes, facts, observations) source, base_language
principle An operational IF/THEN rule derived from a tension origin_tension, if_clause, then_clause, grey_zone, presuppositions
defeated A defeated belief (historical memory) motive, replaced_by
meta_note Reflection on using the system creation_date
presupposition An implicit assumption a principle silently rests on (mapped via the "Map presuppositions" command) presupposes_of

Design invariant: the layer of a note lives exclusively in its frontmatter. Files never move between folders when a layer changes.

Core features

  • Note creation: guided modals for tensions and substrate notes + free-form input with AI classification.
  • Layer transitions (Eleva, Risolvi, Archivia) happen via frontmatter — files never move.
  • Contradiction Hunter (AI): scans open tensions and substrates, identifies contradictory pairs with a confidence rating. Constraint: it identifies, it does not resolve.
  • Antinomia Graph View (Cytoscape.js + fcose) with per-layer clusters, smooth zoom animations, 6 theme presets.
  • Presupposition mapping (AI): for each principle, identify the implicit assumptions it rests on — these become first-class notes you can challenge.
  • Multi-backend AI: Anthropic Cloud, OpenAI, Groq, OpenRouter, LM Studio local, Ollama local. Multiple profiles with a Hunter-specific override.
  • Mobile support: works on Obsidian mobile (desktop + mobile). For local LLMs from mobile, use bridge networking — see Using local LLMs from mobile below.
  • Complete onboarding: Welcome modal, 7-card tutorial, 21-note example vault (with KEY note for measuring the Hunter), getting-started checklist.

Installation

From the Obsidian Community Store (recommended)

  1. In Obsidian: Settings → Community plugins → Browse
  2. Search "Antinomia" and click Install, then Enable

Or open it directly: obsidian://show-plugin?id=antinomia. Updates arrive automatically through Obsidian.

With BRAT (early/beta builds)

BRAT (Beta Reviewers Auto-update Tool) tracks this repo's latest release directly — useful for pre-store builds:

  1. Install BRAT from Obsidian Community plugins
  2. Open Settings → BRAT → Add Beta plugin
  3. Paste this repo URL: https://github.com/demartinogiuseppe/antinomia
  4. Confirm → BRAT fetches the latest release
  5. Settings → Community plugins → enable Antinomia

BRAT will automatically check for new releases and prompt you to update.

From release zip (manual alternative)

  1. Download main.js, manifest.json (and styles.css if present) from the latest release
  2. Copy them into <YOUR_VAULT>/.obsidian/plugins/antinomia/
  3. Settings → Community plugins → reload → enable Antinomia

See BETA-INSTALL.md for the extended beta-tester guide (dedicated vault, Front Matter Title, AI configuration, example vault).

Build from source

npm install
npm run build

This produces a production main.js at the repo root (alongside manifest.json + styles.css). For real-world use, copy those files into <YOUR_VAULT>/.obsidian/plugins/antinomia/.

Prerequisites

  • Obsidian 1.7.2+
  • Front Matter Title community plugin (recommended) — shows human titles instead of timestamp basenames (T-20260601-...) in the File Explorer. Antinomia flags this if missing.
  • At least one AI backend for Hunter / Propose / Map presuppositions:
    • Free local: LM Studio with a loaded model (e.g., Qwen3 14B distill)
    • Paid cloud: Anthropic / OpenAI / Groq / OpenRouter API key

⚠️ Cloud APIs incur per-token cost. If you're cost-sensitive, use LM Studio or Ollama locally (free, private, models downloaded once).

Privacy & network use

Antinomia makes no autonomous network requests and contains no telemetry, analytics, or tracking of any kind. It never "phones home."

  • The network is used only when you explicitly invoke an AI feature (Contradiction Hunter, propose title, propose IF/THEN, map presuppositions, classify, free-form input, PDF / YouTube concept extraction). Each such action sends a single request to the AI backend you configured, and nowhere else.
  • Cloud backends (Anthropic, OpenAI, Groq, OpenRouter): when you run an AI feature, the content of the notes involved in that action (the text the feature needs) is sent to that provider for processing, subject to the provider's own terms and privacy policy. Choose your provider accordingly.
  • Local backends (LM Studio, Ollama): requests go to your local AI server — by default localhost, but Antinomia also recognizes bridge addresses (Tailscale *.ts.net, LAN IPs 10.* / 192.168.* / 172.16-31.*, and .local / .lan / .home / .internal TLDs) as "local" — i.e., privacy-equivalent to localhost. Nothing leaves your devices. This is the privacy-preserving option. See Using local LLMs from mobile for setup.
  • YouTube transcript fetch additionally contacts YouTube's public timedtext endpoint to download captions for a video URL you provide; if that fails it offers an opt-in paste fallback. No video data is sent anywhere else.
  • API keys are stored locally in your vault's .obsidian/plugins/antinomia/data.json (Obsidian's standard plugin-settings file), in plain text. If you sync your vault (Obsidian Sync, iCloud, Git, Dropbox, …), that file — and your keys — travels with it. Keep it out of public repositories and shared folders.

Using local LLMs from mobile

Antinomia works on Obsidian mobile. By default, the mobile app can reach cloud backends (Anthropic, OpenAI, Groq, OpenRouter) directly. Local backends (LM Studio, Ollama) are by default reachable only at localhost, which mobile devices cannot access.

To use your home LM Studio or Ollama from mobile, you need to expose them via a network-reachable address. Three setup paths, in increasing complexity:

Option 1 — Tailscale (recommended)

Tailscale creates a private mesh VPN between your devices. Your phone is "virtually" on the same network as your desktop, even from cellular. End-to-end encrypted, zero firewall configuration, free for personal use.

  1. Install Tailscale on your desktop (where LM Studio / Ollama runs) and on your phone.
  2. Sign in to both with the same account.
  3. On the desktop, find your Tailscale Magic DNS name (e.g., mydesktop.tail1234.ts.net).
  4. In LM Studio (or Ollama), bind the server to 0.0.0.0 (all interfaces) instead of 127.0.0.1.
  5. In Antinomia mobile, set the profile baseUrl to: http://mydesktop.tail1234.ts.net:1234/v1 (LM Studio) or http://mydesktop.tail1234.ts.net:11434/v1 (Ollama).

Your notes never leave your devices — Tailscale is end-to-end encrypted.

Option 2 — Same-WiFi LAN

If both devices are on the same home WiFi:

  1. Find your desktop's local IP (e.g., 192.168.1.50). On Windows: ipconfig. On macOS: System Settings → Network.
  2. In LM Studio / Ollama, bind to 0.0.0.0.
  3. Configure your desktop firewall to allow inbound connections on port 1234 (LM Studio) or 11434 (Ollama) from the local network only.
  4. In Antinomia mobile, set baseUrl to: http://192.168.1.50:1234/v1.

Works only when the phone is on the same WiFi.

Option 3 — Cloudflare Tunnel

For access from anywhere via the public internet, with HTTPS:

  1. Install cloudflared on your desktop.
  2. Run cloudflared tunnel --url http://localhost:1234 (or your local port).
  3. Cloudflare gives you a public HTTPS URL (e.g., https://random-name.trycloudflare.com).
  4. In Antinomia mobile, set baseUrl to that URL + /v1.

Cloudflare proxies the traffic. For long-term use, configure an Argo Tunnel with a custom domain.

Why this matters

These setups let you run AI on your own machine from your phone — your notes go to your desktop's local model, never to a cloud AI provider. Antinomia treats Tailscale and private LAN / internal addresses as Local backend in the UI (privacy notices reflect this). A Cloudflare Tunnel (Option 3) reaches your local model too, but its traffic is proxied through Cloudflare's network, so Antinomia does not classify those public URLs as local.

Philosophy

Antinomia is not a tool to fill up. It is a practice. The vault grows as you encounter contradictions in your own thinking (substrate). Tensions emerge from the material — they are not designed. The Hunter shows you contradictions you hadn't seen — not to resolve them for you, but to force you to think them through.

When you understand a tension well enough to formulate it as an operational principle (IF/THEN/GREY ZONE), you elevate it. When a belief proves false, it goes into the Defeated archive as a memory of what was not true. The graph that emerges is the map of your epistemic history.

Citation

If you use Antinomia in academic or research contexts, please cite both:

The conceptual paper (the idea behind Personal Tension Management):

De Martino, G. (2026). Antinomia: Personal Tension Management. Zenodo. https://doi.org/10.5281/zenodo.20369124

The software (this plugin):

De Martino, G. (2026). Antinomia: an Obsidian plugin for Personal Tension Management [Software]. Zenodo. https://doi.org/10.5281/zenodo.20506815

This is a concept DOI — it always resolves to the latest archived version. See CITATION.cff for the structured format.

Intellectual references

Hegel (dialectics), David Bohm (dialogue that suspends defense), Karl Popper (falsifiability), Thomas Kuhn (scientific revolutions emerge from anomalies), Friedrich Hayek (spontaneous order, local knowledge).

License

MIT. © 2026 Giuseppe De Martino.

Author

Giuseppe De Martino. For discussions, ideas, bug reports: open an issue on GitHub.

Sponsors