Skip to main content

Configuration

💡TL;DR

Notemd configuration covers 6 main areas: (1) LLM providers and API keys, (2) Per-task model selection, (3) Output paths and templates, (4) UI language and output language, (5) Search services (Tavily/DuckDuckGo), (6) Custom workflows. All settings are in Obsidian Settings → Notemd. Changes take effect immediately.

Settings Overview​

Access: Obsidian Settings (⚙️) → Notemd

Main Sections​

  1. Provider Configuration — API keys, base URLs, models
  2. Task Settings — Per-task model selection
  3. Output Configuration — Where to save generated files
  4. Language Settings — UI and output languages
  5. Search Integration — Tavily/DuckDuckGo setup
  6. Workflows — Custom button chains
  7. Advanced — Diagnostics, batch processing, chunking

1. Provider Configuration​

Adding a Provider​

  1. Select provider from dropdown (30+ available)
  2. Enter API Key
  3. Optional: Set custom base URL (for proxies/gateways)
  4. Click "Get Model List" to verify connection
  5. Select default model

Common Providers​

OpenAI​

Provider: OpenAI
API Key: sk-...
Base URL: https://api.openai.com/v1 (default)
Model: gpt-4o-mini

Anthropic (Claude)​

Provider: Anthropic
API Key: sk-ant-...
Model: claude-sonnet-4-6

Ollama (Local)​

Provider: Ollama
Base URL: http://localhost:11434 (default)
Model: llama3.1
API Key: (not required)

DeepSeek​

Provider: DeepSeek
API Key: sk-...
Model: deepseek-chat

Multiple Providers​

You can configure multiple providers and switch between them per task:

  • Main provider: Used as fallback
  • Task-specific providers: Override for specific tasks

2. Task-Specific Models​

Configure different models for different tasks to optimize cost/quality:

Task Types​

TaskDefault ModelPurpose
LinkingMain providerIdentify concepts for wiki-links
Concept GenerationMain providerExtract and describe concepts
ResearchMain providerSummarize web search results
TranslationMain providerTranslate text
Diagram GenerationMain providerGenerate Mermaid/Canvas diagrams

Example: Cost-Optimized Setup​

Linking: gpt-4o-mini (fast, many calls)
Concept Generation: claude-sonnet-4-6 (quality extraction)
Research: gpt-4o (web-grounded)
Translation: gemini-flash (multilingual)
Diagram: ollama/llama3.1 (free, local)

How to Configure​

  1. Settings → Notemd → Task Settings
  2. For each task, choose:
    • ✅ Use main provider (default)
    • 🔄 Use custom provider → select provider + model

3. Output Configuration​

Concept Notes Output​

Where new concept notes are saved:

Settings → Output → Concept Folder
Default: concepts/

Template:

Settings → Output → Concept Template
Default:
---
tags: [concept]
created: {{date}}
---

# {{concept}}

{{description}}

## Backlinks
{{backlinks}}

Folder Structure Options​

Option 1: Flat (all concepts in one folder)

vault/
└── concepts/
    ├── Machine Learning.md
    ├── Neural Networks.md
    └── Deep Learning.md

Option 2: Hierarchical (by source folder)

vault/
├── papers/
│   └── concepts/
│       └── Transformer.md
└── notes/
    └── concepts/
        └── Attention Mechanism.md

Configure: Settings → Output → Use Hierarchical Folders

Diagram Output​

Where diagrams are saved:

Settings → Output → Diagram Folder
Default: diagrams/

Format:

  • .mmd (Mermaid source)
  • .canvas (JSON Canvas)
  • .html (Vega-Lite)

4. Language Settings​

UI Language​

Interface language (buttons, menus, dialogs):

Settings → Language → UI Language
Options: 21+ languages
Default: System language

Supported: English, 简体中文, 繁體中文, 日本語, 한국어, Español, Français, Deutsch, Русский, العربية, and more.

Output Language​

Generated content language (concept notes, research summaries):

Settings → Language → Output Language
Options: Same as UI languages
Default: Same as UI

Use case: Chinese UI + English output for bilingual research.

Translation Target​

Default target language for translation tasks:

Settings → Translation → Target Language
Default: English

5. Search Integration​

AI-optimized search service:

Settings → Search → Service: Tavily
API Key: tvly-...

Get key: tavily.com

Features:

  • High-quality results
  • AI-optimized summaries
  • Source citations

DuckDuckGo (Experimental)​

No API key needed:

Settings → Search → Service: DuckDuckGo

Note: Lower quality, no customization.

6. Workflow Configuration​

Creating Custom Workflows​

Workflows chain multiple tasks into one button.

  1. Settings → Workflows → Add Workflow
  2. Name: "Paper Analysis Flow"
  3. DSL: process-current-add-links>extract-concepts-current>research-and-summarize>summarize-as-mermaid
  4. Save

New button appears in left sidebar.

DSL Syntax​

Format: action1>action2>action3

Available Actions:

  • process-current-add-links — Add wiki-links to current note
  • extract-concepts-current — Generate concept notes
  • research-and-summarize — Web research + summary
  • translate-current — Translate current note
  • summarize-as-mermaid — Generate Mermaid diagram
  • process-folder-add-links — Batch process folder

Example Workflows:

Quick Concept Extraction:

process-current-add-links>extract-concepts-current

Deep Research:

process-current-add-links>research-and-summarize>extract-concepts-current

Multilingual Analysis:

translate-current>process-current-add-links>extract-concepts-current

7. Advanced Settings​

Batch Processing​

Process multiple files in parallel:

Settings → Advanced → Batch Processing
Concurrency: 3 (default)

Higher = faster but more API calls.

Smart Chunking​

For long notes, split into chunks:

Settings → Advanced → Smart Chunking
Chunk Size: 2000 words (default)
Overlap: 200 words (default)

Prevents token limit errors.

Diagnostics Panel​

View detailed logs:

Settings → Advanced → Enable Diagnostics

Shows:

  • API requests/responses
  • Token usage
  • Processing times
  • Errors

Duplicate Concept Detection​

Avoid near-duplicate concepts:

Settings → Advanced → Synonym Suppression
Threshold: 0.8 (default, 0-1)

Higher = more aggressive merging.

Configuration Profiles​

Export/Import Settings​

Share configurations across devices:

  1. Settings → Advanced → Export Configuration
  2. Saves notemd-config.json
  3. On new device: Import Configuration

What's included:

  • Provider API keys (optional)
  • Model selections
  • Output paths
  • Custom workflows

For Students​

Provider: OpenAI (gpt-4o-mini)
Concept Folder: concepts/
UI Language: System
Output Language: System
Workflow: "Study Flow" (add-links>extract-concepts)

For Researchers​

Providers:
  - Linking: DeepSeek (cheap, many calls)
  - Concepts: Claude Sonnet (high quality)
  - Research: GPT-4o (web-grounded)
Concept Folder: notes/concepts/
Hierarchical: Yes
Workflow: "Paper Flow" (add-links>extract>research>diagram)

For Privacy-Conscious Users​

Provider: Ollama (llama3.1)
All tasks: Local model
Search: DuckDuckGo (no API)
Diagnostics: Off

For Bilingual Work​

UI Language: Chinese
Output Language: English
Translation Target: Chinese
Provider: Gemini (multilingual)

Troubleshooting Configuration​

Settings Not Saved​

  • Check file permissions in .obsidian/plugins/notemd/
  • Try: Disable plugin → Re-enable

API Key Invalid​

  • No extra spaces
  • Check expiration
  • Verify credits in provider dashboard

Models Not Loading​

  • Click "Get Model List" to refresh
  • Check network connectivity
  • Try different provider

Next Steps​