🔥 Calcifer

AI-Powered Assistant for Obsidian

Your intelligent AI companion that understands your vault.

RAG-powered chat · Smart auto-tagging · Tool calling · Semantic search · Persistent memory

Features · Installation · Configuration · Commands · Privacy

---

Features

AI Chat with Vault Context Chat interface in the right sidebar Retrieves relevant notes as context for answers Shows sources for every response Supports Ollama and OpenAI-compatible APIs	Smart Auto-Tagging Suggests tags based on note content Uses existing vault tags as reference Configurable auto-apply or suggest-only mode
Tool Calling Create, move, rename, and delete notes Create folders and organize vault structure Configurable confirmation for destructive actions	Memory System Remembers facts about you across conversations Stored locally in plugin data (not sent to cloud) Manageable through settings modal
Note Organization Suggests appropriate folders for notes Based on content similarity to existing notes LLM-enhanced folder recommendations	Semantic Search Full vault indexing with embeddings Find notes by meaning, not just keywords Automatic re-indexing on file changes

---

Screenshots

Auto Tag Suggestions

Tool Calling

---

Requirements

• Obsidian v1.0.0 or higher
• AI API endpoint (one of the following):

Provider	Description
Ollama	Local or remote — complete privacy
OpenAI	Or compatible API (Azure OpenAI, etc.)

---

Installation

From Community Plugins (Coming Soon)

1. Open Settings → Community plugins
2. Search for "Calcifer"
3. Click Install, then Enable

Manual Installation

# 1. Download from the latest release:
#    main.js, manifest.json, styles.css

# 2. Create the plugin folder
mkdir -p <vault>/.obsidian/plugins/calcifer/

# 3. Copy files and enable in Settings → Community plugins

---

Configuration

1. Add an API Endpoint

1. Open Settings → Calcifer
2. Click "Add Ollama" or "Add OpenAI"
3. Configure the endpoint:

Ollama Configuration

Base URL:        http://localhost:11434
Chat Model:      llama3.2
Embedding Model: nomic-embed-text

OpenAI Configuration

Base URL:        https://api.openai.com
API Key:         sk-...
Chat Model:      gpt-4o-mini
Embedding Model: text-embedding-3-small

4. Click "Test" to verify connection
5. Enable the endpoint

2. Index Your Vault

Use command: Calcifer: Re-index Vault
Or enable automatic background indexing in settings.

3. Start Chatting

• Click the bot icon in the left ribbon
• Or use command: Calcifer: Open Chat

---

Commands

Command	Description
Open Chat	Open the chat sidebar
Re-index Vault	Rebuild the embedding index
Stop Indexing	Stop the current indexing process
Clear Embedding Index	Delete all embeddings
Index Current File	Index only the active file
Show Status	Display indexing stats and provider health
Show Memories	Open the memories management modal
Suggest Tags for Current Note	Get AI tag suggestions
Suggest Folder for Current Note	Get folder placement suggestions

---

Settings Reference

Embedding Settings

Setting	Default	Description
Enable Embedding	false	Toggle automatic indexing (enable after configuring provider)
Batch Size	1	Concurrent embedding requests
Chunk Size	1000	Characters per text chunk
Chunk Overlap	200	Overlap between chunks
Debounce Delay	5000	Milliseconds to wait before indexing changed files
Exclude Patterns	templates/**	Glob patterns to skip

RAG Settings

Setting	Default	Description
Top K Results	5	Context chunks to retrieve
Minimum Score	0.5	Similarity threshold (0-1)
Include Frontmatter	true	Add metadata to context
Max Context Length	8000	Total context character limit

Chat Settings

Setting	Default	Description
System Prompt	(built-in)	Customize assistant behavior
Include Chat History	true	Send previous messages
Max History Messages	10	History limit
Temperature	0.7	Response creativity (0-2)
Max Tokens	2048	Response length limit

Tool Calling Settings

Setting	Default	Description
Enable Tool Calling	true	Allow AI to perform vault operations
Require Confirmation	false	Ask before executing destructive tools

Memory Settings

Setting	Default	Description
Enable Memory	true	Store persistent memories
Max Memories	100	Storage limit
Include in Context	true	Send memories with queries

Auto-Tagging Settings

Setting	Default	Description
Enable Auto-Tag	false	Activate tagging feature (opt-in)
Mode	suggest	auto (apply) or suggest (show modal)
Max Suggestions	5	Tags per note
Use Existing Tags	true	Prefer vault tags
Confidence Threshold	0.8	Auto-apply threshold (0-1)

Organization Settings

Setting	Default	Description
Enable Auto-Organize	true	Activate folder suggestions
Mode	suggest	auto (move) or suggest (ask)
Confidence Threshold	0.9	Auto-move threshold (0-1)

Performance Settings

Setting	Default	Description
Enable on Mobile	true	Run on mobile devices
Rate Limit (RPM)	60	API requests per minute
Request Timeout	120	Seconds before timeout
Use Native Fetch	false	Use native fetch for internal CAs

UI Settings

Setting	Default	Description
Show Context Sources	true	Display sources in chat responses
Show Indexing Progress	true	Show indexing notifications

---

Mobile Support

Calcifer is fully functional on mobile devices:

• Chat interface optimized for touch
• Background indexing respects mobile resources
• All features work offline with local Ollama

---

Privacy & Security

Aspect	Details
Local Processing	All embeddings stored locally in IndexedDB
No Cloud Storage	Plugin data never leaves your device
API Choice	Use local Ollama for complete privacy
Memory Control	View and delete any stored memories

---

Network Usage Disclosure

Important: No data is sent to any server until you configure an API endpoint.

Service	Purpose	Data Sent
Ollama (local/remote)	Chat completions, embeddings	Note content for context, user messages
OpenAI (or compatible)	Chat completions, embeddings	Note content for context, user messages

• You control which provider to use (local Ollama = no external network)
• Note content is sent as context for AI responses (chunks of ~1000 chars)
• No telemetry or analytics are collected by this plugin

---

Development

# Clone the repository
git clone https://github.com/anyesh/obsidian-calcifer.git
cd obsidian-calcifer

# Install dependencies
npm install

# Development mode (watch)
npm run dev

# Production build
npm run build

---

Troubleshooting

"No provider configured"

• Add at least one API endpoint in settings
• Ensure the endpoint is enabled
• Test the connection

"Connection failed"

• Check if Ollama is running (ollama serve)
• Verify the base URL is correct
• Check firewall/network settings

Indexing is slow

• Reduce batch size for limited resources
• Exclude large folders (templates, archives)
• Mobile devices may need smaller chunk sizes

Chat responses are irrelevant

• Ensure vault is indexed (check status bar)
• Lower the minimum score threshold
• Increase Top K for more context

---

License

MIT License — see LICENSE for details.

---

Acknowledgments

Made for the Obsidian community

---

Back to Top