Dualyze Notes

approved

by DualyzeAI

Find similar notes and create safe merge drafts in Obsidian. - This plugin has not been manually reviewed by Obsidian staff.

↓ 241 downloadsUpdated 1mo agoMIT

Install in Obsidian View on GitHub

Dualyze Notes

Find overlapping notes, compare them side by side, and create safe merge drafts in Obsidian.

Obsidian vaults naturally accumulate similar notes: drafts, copied ideas, research fragments, and near-duplicates.

Dualyze Notes scans your vault for overlap in titles, headings, tags, links, and content. It helps you compare notes, understand what they share, identify what is unique, and create merge drafts without modifying the originals.

Dualyze Notes uses local similarity scoring. It does not send your notes to an external AI service.

📖 Step-by-step Tutorial

See how Alice discovers that two habit notes are 78% similar and safely generates a Merge Draft.

Learn how to:

Find overlapping notes
Compare similarities
Preserve originals
Create merge drafts

→ Read the complete tutorial: How Alice Discovered Her Two Habit Notes Were 78% the Same

When to use Dualyze Notes

Find duplicate or near-duplicate notes
Compare an old draft with a newer version
Review similar research notes before consolidating
Merge scattered ideas safely
Clean up a growing vault without losing unique content

Features

Find similar notes — scan the current folder or entire vault and display results in a persistent side panel
Compare notes side by side — see a score breakdown across five dimensions with common and unique keyword chips
Understand why notes match — shared headings, tags, links, and keywords shown at a glance
Create safe merge drafts — generate a merge scaffold in Dualyze Merge Drafts/; originals are never modified
Save comparison reports — export a structured Markdown comparison to Dualyze Reports/
Tune similarity settings — adjust threshold, scan scope, and per-dimension weights to match your workflow

Installation

Community plugins (recommended)

Open Settings → Community plugins → Browse
Search for Dualyze Notes
Install and enable

Manual

Download main.js, styles.css, and manifest.json from the latest release and copy them to .obsidian/plugins/dualyze-notes/ in your vault.

Tutorial: try it with the sample vault

The repository includes a ready-made vault in sample-vault/. It contains two topic folders — pasta recipes and travel plans — designed to show the full range of similarity labels. No setup is needed — just open the folder and follow the steps below.

Step 1 — Open the sample vault in Obsidian

In Obsidian, choose Open folder as vault and select the sample-vault/ folder from this repository. You will see:

sample-vault/
  Start Here.md
  recipes/
    Spaghetti Carbonara.md
    Spaghetti Carbonara (Draft).md
    Spaghetti Gricia.md
    Cacio e Pepe.md
  travel/
    Tokyo Travel Plan.md
    Tokyo Travel Plan (Draft).md
    Osaka Travel Plan.md
    Kyoto Travel Plan.md
    Kyoto Day Trip.md
    Seoul Travel Notes.md

Make sure the Dualyze Notes plugin is installed and enabled before continuing.

Step 2 — Find similar notes

Open Spaghetti Carbonara.md from the recipes/ folder. Then do one of the following:

Right-click the note tab or the file in the explorer → Find similar notes
Open the command palette (Ctrl/Cmd + P) → Find similar notes

The Similar Notes panel opens on the right side. At the default threshold of 70%, only notes that score 70% or above are shown. The Draft scores ~81%, so it appears immediately.

Simplified text representation:

Similar notes
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Spaghetti Carbonara
[Rescan]  Threshold: ●━━━━━━━━  70%
───────────────────────────────────
~81%  Merge candidate
Spaghetti Carbonara (Draft)
[Compare]  [Open]  [Report]  [Create merge draft]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Now lower the threshold slider to about 30%. The other two notes come into view:

Simplified text representation:

Similar notes
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Spaghetti Carbonara
[Rescan]  Threshold: ━━━━━━━━●━  30%
───────────────────────────────────
~81%  Merge candidate
Spaghetti Carbonara (Draft)
[Compare]  [Open]  [Report]  [Create merge draft]

~50%  Related
Spaghetti Gricia
[Compare]  [Open]  [Report]  [Create merge draft]

~33%  Weakly related
Cacio e Pepe
[Compare]  [Open]  [Report]  [Create merge draft]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Exact scores vary slightly depending on your settings. The labels (Merge candidate, Related, Weakly related) are the reliable signal.

The threshold slider is a real-time filter — it does not rescan. Drag it to narrow or widen the results without reanalysing the notes.

Step 3 — Compare two notes

Click Compare on the Spaghetti Carbonara (Draft) card. The Compare Notes modal opens.

Simplified text representation:

Spaghetti Carbonara  vs  Spaghetti Carbonara (Draft)

~81%   [Merge candidate]

Title      ██████████████████████░░░░░░░  ~67%
Headings   █████████████████████████████  100%
Content    ████████████████████░░░░░░░░░  ~62%
Tags       ██████████████████████████░░░  ~75%
Links      █████████████████████████████  100%

Common
  Headings: Ingredients  Method  Tips  See also
  Tags: pasta  italian  roman
  Keywords: spaghetti  guanciale  eggs  Pecorino  Romano  pasta  water  black  pepper

Unique to Spaghetti Carbonara      Unique to Spaghetti Carbonara (Draft)
  Keywords: cream  authentic          Keywords: rigatoni  parmigiano

Reading this modal:

Score bars — Title scores ~67% because both titles share "spaghetti" and "carbonara" but the Draft filename has an extra word. Headings score 100% because all headings now match exactly. Links score 100% because both notes point to the same two notes. Content scores ~62% because the two notes describe the same dish in similar but not identical words.
Common section — shows exactly what the two notes already agree on.
Unique section — shows keywords found only in each note. Use this to spot content that the draft will not capture automatically. When headings are 100%, no unique headings appear here.

Step 4 — Create a draft

Click Create merge draft at the bottom of the modal. Dualyze Notes creates a new note in Dualyze Merge Drafts/ and opens it automatically. The original notes are untouched.

The draft contains:

Links back to both source notes
Each shared heading with its content pulled from both notes
Common keywords listed for reference

For each shared heading, the draft shows:

Identical content — shown once
Different content — both versions shown side by side, labelled with the source note name
Missing from one note — only the available version is shown

Use this as your working document. Decide which version to keep for each section, then decide whether to keep, archive, or delete the originals once you are satisfied.

Step 5 — Create a report

Go back to the panel and click Report on the Spaghetti Gricia card. A report note is saved in Dualyze Reports/ and opened. Reports are permanent notes — link to them from your own notes, include them in a review workflow, or just delete them when done.

Step 6 — Compare with a note from a different folder

Use the command palette → Compare current note with another note. A search modal appears. Type the name of any note in the vault, select it, and the same Compare modal opens — regardless of which folder the note is in.

Try the travel folder

The same steps work with the travel/ folder. Open Tokyo Travel Plan.md and repeat the scan. At 70% threshold you will see the Draft. Lower the threshold to about 30% to see all five results:

Note	Expected label
Tokyo Travel Plan (Draft)	Merge candidate (~84%)
Osaka Travel Plan	Related (~54%)
Kyoto Travel Plan	Related (~54%)
Kyoto Day Trip	Weakly related (~33%)
Seoul Travel Notes	Not similar

Notes from different folders do not affect each other. Scan scope defaults to the current folder, so scanning inside recipes/ only compares recipe notes, and scanning inside travel/ only compares travel notes.

Usage

Finding similar notes

There are two ways to start a scan:

Right-click menu Right-click any note in the file explorer or editor tab and choose Find similar notes. The Similar Notes panel opens on the right side and immediately shows results for that note.

Command palette Open the command palette (Ctrl/Cmd + P), search for Find similar notes, and run it while a Markdown note is active.

Example

See a real-world comparison workflow:

How Alice Discovered Her Two Habit Notes Were 78% the Same

Similar Notes panel

The panel stays open alongside your notes. It shows all notes in the same folder that meet the similarity threshold, sorted by score from highest to lowest.

Simplified text representation:

┌─────────────────────────────────────────┐
│ Spaghetti Carbonara                     │  ← source note
│ [Rescan]  Threshold: [━━━━●━━━] 70%    │
├─────────────────────────────────────────┤
│ 76%  Merge candidate                    │
│ Spaghetti Carbonara (Draft)             │
│ [Compare] [Open] [Report] [Create merge draft] │
├─────────────────────────────────────────┤
│ 54%  Related                            │
│ Pasta Basics                            │
│ [Compare] [Open] [Report] [Create merge draft] │
└─────────────────────────────────────────┘

Threshold slider — drag the slider to raise or lower the minimum score displayed. Changes take effect immediately without rescanning. The default is 70%.

Rescan — re-runs the scan against the current folder. Use this after editing notes.

Per-card actions:

Button	What it does
Compare	Opens the Compare Notes modal for a detailed breakdown
Open	Opens the target note in a new tab
Report	Saves a comparison report to `Dualyze Reports/` and opens it
Create merge draft	Saves a merge draft to `Dualyze Merge Drafts/` and opens it

Comparing two notes

From the Similar Notes panel: click the Compare button on any result card.

From the command palette: search for Compare current note with another note. A search modal appears — type to filter and select the second note.

From the right-click menu: right-click a note and choose Compare with....

Compare Notes modal

The modal shows the full breakdown of how two notes relate to each other.

Simplified text representation:

┌──────────────────────────────────────────────────────┐
│  Spaghetti Carbonara  vs  Spaghetti Carbonara (Draft)│
│                                                      │
│  76%   [Merge candidate]                             │
│                                                      │
│  Title    ████████████████████████████░░░░░  85%    │
│  Headings ████████████████░░░░░░░░░░░░░░░░  60%    │
│  Content  ████████████░░░░░░░░░░░░░░░░░░░░  45%    │
│  Tags     ████████████████████░░░░░░░░░░░░  70%    │
│  Links    ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░   0%    │
│                                                      │
│  Common                                              │
│  Headings: `Ingredients` `Method`                   │
│  Tags: `pasta` `italian`                             │
│  Keywords: `carbonara` `egg` `guanciale`             │
│                                                      │
│  Unique to Note A         Unique to Note B           │
│  Headings: `History`      Headings: `Variations`     │
│  Keywords: `origin`       Keywords: `cream`          │
│                                                      │
│  Original notes will not be modified.                │
│  [Report]  [Create merge draft]                      │
└──────────────────────────────────────────────────────┘

Score bars — five horizontal bars show per-dimension scores. The overall score is the weighted sum.

Badge — a label summarises the relationship at a glance:

Score	Badge	Meaning
90% and above	Duplicate	Near-identical notes; likely safe to merge
70–89%	Merge candidate	Significant overlap; worth reviewing
50–69%	Related	Partially related; may share a theme
30–49%	Weakly related	Minor overlap
Below 30%	Not similar	Essentially unrelated

Common section — headings, tags, links, and keywords that appear in both notes.

Unique section — headings and keywords found only in each respective note. Use this to spot what would be lost if one note were deleted.

Creating a report

Click Report (in the panel or modal). A Markdown file is created in Dualyze Reports/ and opened automatically. The report contains:

Overall score and label
Score breakdown table (Title / Headings / Tags / Links / Content)
Common headings, tags, links, and keywords
Unique headings and keywords per note
Wiki-links back to both source notes

The report folder is configurable in Settings. Reports are standalone notes — you can keep, link, or delete them freely.

Creating a merge draft

Click Create merge draft (in the panel or modal). A new note is created in Dualyze Merge Drafts/ and opened automatically.

The draft contains:

Links to both source notes
Each shared heading with its content pulled from both notes
Common keywords listed for reference

For each shared heading, the content is resolved as follows:

Situation	What appears in the draft
Both notes have identical content	Shown once
Both notes have different content	Both versions shown with `> From Note Name` labels
Only one note has content	That version shown without a label

The original notes are not modified. The draft is a new file; deleting it has no effect on the sources.

Learn by Example

See how Merge Drafts preserve original notes while combining related knowledge:

How Alice Discovered Her Two Habit Notes Were 78% the Same

Settings

Open Settings → Dualyze Notes to configure the plugin.

Basic

Setting	Default	Description
Similarity threshold	0.70	Minimum score (0–1) for a note to appear as similar. Notes below this value are hidden.
Max results	10	Maximum number of similar notes shown in the panel per scan.
Scan scope	Current folder	Where to look for candidates. `Current folder` scans only notes in the same folder as the source. `Entire vault` searches everywhere (slower on large vaults).

Output folders

Setting	Default	Description
Comparison report folder	`Dualyze Reports`	Where report notes are saved. Created automatically if it does not exist.
Merge draft folder	`Dualyze Merge Drafts`	Where draft notes are saved. Created automatically if it does not exist.
Archive folder	`_archive/dualyze-notes`	Where notes are moved when using the archive action (disabled by default).

Exclude

Exclude folders — folder names to skip during scanning, one per line. Default list: templates, _archive, .trash, node_modules, attachments.

Exclude file patterns — glob-style patterns for files to skip, one per line. Default: *.canvas, *.excalidraw.md.

Similarity weights

The overall score is a weighted average of five dimensions. All five weights should add up to 1.0 (100%). A warning appears in Settings if they do not.

Dimension	Default weight	What it measures
Title	30%	Overlap between the two note titles using character n-grams
Headings	25%	Jaccard similarity of the heading lists
Tags	15%	Jaccard similarity of the tag lists
Links	15%	Jaccard similarity of internal wiki-links
Content	15%	Token overlap across the note body

Increase Title weight to find duplicates with similar names more aggressively. Increase Content weight to surface notes that cover the same material regardless of how they are titled.

Markdown parsing

Setting	Default	Description
Include code blocks	Off	When off, content inside fenced code blocks is excluded from similarity analysis. Turn on if your notes contain meaningful code you want compared.
Include frontmatter in content	Off	When off, YAML frontmatter is not counted as body content. Turn on if your frontmatter contains topic-relevant text.
N-gram size	2	Character n-gram size used when tokenising text. The default of 2 works well for CJK text and mixed-language vaults. Lower values make matching more permissive; higher values make it stricter.

Safety

Setting	Default	Description
Enable archive action	Off	Adds an Archive button to result cards. Clicking it moves the note to the archive folder. Off by default so notes cannot be moved accidentally.
Confirm before moving files	On	Shows a confirmation dialog before any file move. Recommended to leave on.

How similarity is calculated

Dualyze Notes works entirely locally inside your vault. No notes are sent to an external service.

Each note is parsed into five feature sets: title tokens, heading list, tag list, link list, and body tokens.

For structured lists (headings, tags, links), the score is the Jaccard similarity: the size of the intersection divided by the size of the union.

For body content, token overlap is measured using a normalised intersection count.

Title matching uses character n-grams rather than whole words, so partial-word overlaps (useful for CJK text) are captured.

The five dimension scores are combined as a weighted sum to produce the final overall score (0.0–1.0).

Documentation

Tutorials

Related tools

Dualyze Notes is part of the DualyzeAI workflow.

DualyzeAI — compare two websites with AI and save structured analysis to Obsidian
AI Context Pack — turn organised notes into AI-ready context packs for ChatGPT, Claude, Gemini, NotebookLM, and more

License

MIT

dualyzeAI

For plugin developers

Search results and similarity scores are powered by semantic analysis of your plugin's README. If your plugin isn't appearing for searches you'd expect, try updating your README to clearly describe your plugin's purpose, features, and use cases.