Configuration
Complete reference for all Claudian settings. Access settings via Settings > Claudian.
| Setting | Description |
|---|---|
| Name | Language |
| Description | Interface language for the plugin |
| Default | Auto (follows Obsidian's language) |
| Options | Auto, English, Español, 中文, Deutsch, Français, 日本語 |
The plugin automatically detects Obsidian's language setting, or you can manually override it.
| Setting | Description |
|---|---|
| Name | API Key |
| Description | Your Anthropic API key for Claude |
| Default | Empty |
| Required | Yes |
Get your API key from console.anthropic.com. The key is stored locally in your vault's plugin data file and is never sent to external servers (except Anthropic's API).
| Setting | Description |
|---|---|
| Name | Execution Mode |
| Description | How Claudian selects the optimal model for each task |
| Default | Automatic |
| Options | Automatic, Economic, Maximum Quality |
Mode Comparison:
| Mode | Description | Best For |
|---|---|---|
| Automatic | Haiku analyzes each task and routes to the optimal model | Most users - balances cost and quality |
| Economic | All tasks use Haiku | Budget-conscious users, simple tasks |
| Maximum Quality | All tasks use Opus | Complex reasoning, critical work |
How Automatic Mode Works:
The Model Orchestrator uses Haiku (fast and cheap) to classify each task's complexity:
| Complexity | Model Used | Example Tasks |
|---|---|---|
| Simple | Haiku 4.5 | List files, copy, move, delete, placeholder content |
| Moderate | Sonnet 4 | Write content, summarize, translate, explain |
| Complex | Sonnet 4 | Multi-file operations, batch processing, refactoring |
| Deep | Opus 4 | Analysis, strategic planning, knowledge synthesis |
This intelligent routing optimizes costs while ensuring quality where it matters.
| Setting | Description |
|---|---|
| Name | Notes folder |
| Description | Destination folder for generated notes |
| Default | Claude Notes |
Notes created from chat responses will be saved in this folder. The folder will be created automatically if it doesn't exist.
| Setting | Description |
|---|---|
| Name | Max tokens |
| Description | Maximum tokens in Claude's responses |
| Default | 4096 |
| Range | 1000 - 8192 |
Higher values allow longer responses but may increase API costs.
| Setting | Description |
|---|---|
| Name | Custom instructions |
| Description | Additional instructions for Claude |
| Default | Empty |
Add your own instructions to customize Claude's behavior. These instructions are appended to the built-in system prompt, so core functionality is always preserved. Click Clear to remove all custom instructions.
Examples of custom instructions:
- "Always respond in formal English"
- "Prefer using bullet points over paragraphs"
- "When creating notes, always add a 'status: draft' field"
Note: You cannot override Claude's core identity or base functionality. Custom instructions supplement the default behavior.
These settings control how notes are analyzed when using the "Process active note" command.
| Setting | Description |
|---|---|
| Name | Notes in context |
| Description | Number of note titles to include for wikilink suggestions |
| Default | 100 |
| Range | 10 - 500 |
Higher values provide more potential wikilink targets but increase API token usage.
| Setting | Description |
|---|---|
| Name | Tags in context |
| Description | Number of existing tags to include for tag suggestions |
| Default | 50 |
| Range | 10 - 200 |
Includes your vault's existing tags so Claude can suggest relevant ones.
Settings for natural language vault management.
| Setting | Description |
|---|---|
| Name | Enable agent mode by default |
| Description | Start new chats with agent mode enabled |
| Default | Off |
When enabled, new chat sessions will have agent mode active. You can always toggle it in the chat header.
| Setting | Description |
|---|---|
| Name | Confirm destructive actions |
| Description | Show confirmation dialog before delete operations |
| Default | On |
Strongly recommended to keep enabled. This prevents accidental deletion of notes or folders.
| Setting | Description |
|---|---|
| Name | Protected folders |
| Description | Folders that cannot be modified by agent mode |
| Default | .obsidian, templates, _templates |
Comma-separated list of folder paths. Agent mode will refuse to modify files in these folders.
| Setting | Description |
|---|---|
| Name | Max actions per message |
| Description | Maximum vault operations per agent response |
| Default | 25 |
| Range | 1 - 50 |
Limits the number of actions Claude can execute in a single response to prevent runaway operations.
| Setting | Description |
|---|---|
| Name | Auto-continue |
| Description | Automatically continue truncated responses |
| Default | On |
When enabled, if Claude's response is cut off due to token limits, the plugin will automatically request a continuation. This ensures complete responses for complex tasks.
| Setting | Description |
|---|---|
| Name | Auto-plan |
| Description | Automatically decompose complex tasks into steps |
| Default | On |
When enabled, Claude will break down complex requests into a series of manageable steps before execution. This improves reliability for multi-step operations.
| Setting | Description |
|---|---|
| Name | Context reinforcement |
| Description | Reinforce agent context in long conversations |
| Default | On |
When enabled, the agent's system context is periodically reinforced during long conversations to maintain consistent behavior and prevent context drift.
Settings for monitoring API token usage.
| Setting | Description |
|---|---|
| Name | Show token indicator |
| Description | Display token usage in chat footer |
| Default | On |
When enabled, shows the current session's token usage in the chat footer. Click the indicator to view detailed usage history.
Token indicator shows:
- Input tokens (sent to Claude)
- Output tokens (received from Claude)
- Session totals
- Click to expand usage history modal
Settings for automatic conversation optimization to reduce token usage.
| Setting | Description |
|---|---|
| Name | Automatic context management |
| Description | Automatically summarize conversation history when it gets long |
| Default | On |
When enabled, Claudian automatically summarizes older messages when conversations become long. This reduces token usage while preserving important context, allowing for longer conversations without hitting token limits.
How it works:
- Monitors conversation length
- When threshold is reached, older messages are summarized
- Recent messages remain in full detail
- Summary context is included in new requests
| Setting | Description |
|---|---|
| Name | Summarize after messages |
| Description | Number of messages before triggering automatic summarization |
| Default | 20 |
| Range | 10 - 50 |
Lower values summarize more frequently (more token savings, less context). Higher values keep more full messages before summarizing.
| Setting | Description |
|---|---|
| Name | Max active messages |
| Description | Maximum messages to keep in active context after summarization |
| Default | 50 |
| Range | 20 - 100 |
After summarization, this many recent messages are kept in full detail. Older messages are compressed into a summary.
Token savings estimates:
| Conversation Length | Without Summarization | With Summarization | Savings |
|---|---|---|---|
| 30 messages | ~7,500 tokens | ~3,000 tokens | 60% |
| 50 messages | ~12,500 tokens | ~3,500 tokens | 72% |
| 100 messages | ~25,000 tokens | ~4,000 tokens | 84% |
All settings are stored locally in your vault at:
.obsidian/plugins/claudian/data.json
This file contains your configuration including the API key. Do not share this file publicly.
- Execution Mode: Automatic
- Max tokens: 4096
- Confirm destructive actions: On
- Execution Mode: Automatic
- Notes in context: 200-300
- Tags in context: 100
- Execution Mode: Economic
- Max tokens: 2048
- Execution Mode: Maximum Quality
- Max tokens: 8192
