General Settings - SystemSculpt Documentation | Documentation

Configure SystemSculpt exactly how you want it - from basic setup to advanced customization

Get started in 2 minutes

License or API key: Settings → Enter key → Save
Pick your model: Chat Settings → Default Model
Ready! Click robot icon 🤖 to start chatting

That's the essentials. Dive deeper when you need to.

Settings overview

Tab	What's there	When to use
Setup & Connect	License, providers, directories	Initial setup
Chat Settings	Models, prompts, MCP tools	Customize AI behavior
Embeddings	Semantic search	Find related notes
Advanced	Debug, performance, reset	Troubleshooting

Setup & Connect - The basics

License key

Have a license?

Settings → License Key → Paste key → Save
✓ Unlocks all premium features instantly

Using your own API keys?

Add Provider → Select type → Enter key → Test

Directory setup

SystemSculpt organizes files automatically:

Directory	Default location	What goes there
Chats	`SystemSculpt/Chats`	All conversations
Recordings	`SystemSculpt/Recordings`	Audio files
Attachments	`SystemSculpt/Attachments`	Images in chats
Extractions	`SystemSculpt/Extractions`	Processed PDFs
System Prompts	`SystemSculpt/System Prompts`	Custom prompts

Change locations: Just type new path - folders created automatically

Chat Settings - Your AI preferences

Default model & prompt

Set defaults for new chats:

Change Default Presets → Select model & prompt → Save

System prompt options:

General Use - Balanced, helpful assistant
Concise - Brief, direct responses
Custom - Your own prompt file or text

Model management

Favorite models (star them for quick access):

Model selector → Click ⭐ → Always appears at top

Sort order:

Default - Natural provider order
Alphabetical - By provider then model name

MCP Tools (Agent Mode)

Let AI work with your vault:

Enable MCP Tools → Accept consent → Configure permissions

Available tools:

Tool	What it does	Default permission
read_files	Read vault content	Requires approval
write_file	Create/update files	Requires approval
find_files	Search by name	Auto-approved
grep_vault	Search content	Auto-approved

Pro tip: Start with "requires approval" until you trust the workflow

Provider Settings - Multiple AI sources

Add custom providers

Quick setup for common providers:

Provider	Get API key from	Special notes
OpenAI	platform.openai.com	Various GPT models
Anthropic	console.anthropic.com	Various Claude models
OpenRouter	openrouter.ai	Many models
Ollama	Run locally	No API key needed

Provider configuration

Add Provider → Fill details:
- Name: "Work OpenAI" (your label)
- API Key: sk-...
- Endpoint: (auto-filled usually)
- Test Connection → Save

Multiple providers? Name them clearly: "Personal GPT", "Work Claude"

Advanced Settings - Power tools

Developer options

Debug mode:

Development Mode: ON
→ Console logs, API calls, performance metrics

Performance indicator:

Show in status bar → Monitor memory/lag

Troubleshooting tools

Tool	What it does	When to use
Reset to Factory	Clear all settings	Major issues
Restore Backup	Load previous config	Undo changes
Copy Logs	Export for support	Getting help
Device Info	Show capabilities	Platform issues

Backup system

Auto-backs up daily
Last 7 days retained
Stored in .obsidian/plugins/systemsculpt-ai/.systemsculpt/

Embeddings - Semantic search

Enable smart search

Embeddings tab → Enable → Process

What you get:

Find notes by meaning, not just keywords using OpenAI embeddings
Privacy-first by default: Data sent via the API is not used to train OpenAI models
See related content automatically
Discover hidden connections

Processing time:

Small vault (<500 notes): 5-15 min
Medium (500-2000): 15-60 min
Large (2000+): 1-3 hours

Manage embeddings

Action	What it does	When to use
Process	Generate/update vectors	After major changes
Clear	Remove all embeddings	Switching providers
Stats	View progress	Check completion

Keyboard shortcuts

Essential shortcuts

Action	Default	Customize in
Open Chat	`Ctrl/Cmd + Shift + A`	Settings → Hotkeys
Start Recording	`Ctrl/Cmd + Shift + R`	Search "SystemSculpt"
Add Context	`@` in chat	Built-in
Templates	`/` in chat	Built-in

Chat shortcuts

Send: Enter
New line: Shift + Enter
Cancel AI: Escape
Font size: Ctrl/Cmd + .

Best practices

Initial setup

✅ DO:

Test connection after adding providers
Set reasonable default model (balance cost/quality)
Organize directories by project if needed
Enable embeddings for semantic search

❌ DON'T:

Share API keys
Auto-approve all MCP tools immediately
Ignore backup system
Skip provider testing

Optimization tips

For performance:

Use smaller models for simple tasks
Disable unused features
Clear old chat histories periodically
Monitor with performance indicator

For organization:

Name providers clearly
Create project-specific directories
Backup before major changes
Document custom settings

Common configurations

For researchers

Yaml
Default Model: Balanced model
System Prompt: General Use
Embeddings: Enabled
MCP Tools: Read-only
Directories: Organized by project

For developers

Yaml
Default Model: Advanced model
System Prompt: Concise
Debug Mode: ON
MCP Tools: Full access
Performance Monitor: Visible

For writers

Yaml
Default Model: Premium model
System Prompt: Custom creative
Font Size: Large
Embeddings: Enabled
Audio Recording: Quick access

Troubleshooting

Issue	Quick fix
Connection failed	Check API key, test provider
Settings not saving	Check permissions, try backup
High memory	Disable embeddings temporarily
Slow performance	Use smaller models, check debug

Get help

Settings location: .obsidian/plugins/systemsculpt-ai/data.json
Discord: Join community
Logs: Advanced → Copy Logs for Support

💡 Pro tip: Start simple, add features as needed. Most users only need license key + default model!