Settings & Security
This guide covers how to manage all of ProjectHub's settings and explains the security system.
Accessing Settings
You can access settings through the following methods:
| Method | Description |
|---|---|
| Project Settings button | Gear icon in the project detail header |
| Network Status click | Network indicator on the left side of the toolbar |
Intelligence Settings (AI Configuration)
These are the core settings for configuring and managing AI providers.
BYOK (Bring Your Own Key)
ProjectHub operates on a BYOK model. It uses your own API keys directly. ProjectHub does not store, proxy, or cache any API credentials or responses. All API usage is billed directly to your account with each AI provider.
A BYOK notice badge is displayed at the top of the settings screen.
Screenshot: Full Intelligence settings tab
AI Provider Management
| Provider | Description | Installation Check | Configuration |
|---|---|---|---|
| Claude CLI | Anthropic Claude | claude --version | Model selection, enable/disable |
| Gemini CLI | Google Gemini | gemini --version | Model selection, enable/disable |
| Ollama | Local AI | ollama --version | Host, model, enable/disable |
| OpenCode | Open source | opencode --version | Enable/disable |
For each provider:
- Enable/Disable toggle
- Availability auto-detection (installation status check)
- Model selection (where applicable)
Autonomous Mode
Screenshot: Autonomous mode toggle with description
| Setting | Description |
|---|---|
| Autonomous Mode | Auto-approve AI file operations |
- AI Agent (Vibe Coding): Auto-approves file creation and modification
- Alliance: Auto-selects at Human Checkpoints
- Alliance: Auto-retries at Fix Checkpoints
Caution: In autonomous mode, the AI modifies files without approval. Always create a snapshot first when working on important projects.
Ollama Remote Configuration
You can run Ollama on another computer and connect to it over the network:
Screenshot: Ollama remote configuration fields
Format: host:port|model
Example: 192.168.1.100:11434|llama3
| Field | Description | Example |
|---|---|---|
| Host | Ollama server IP | 192.168.1.100 |
| Port | Ollama server port | 11434 (default) |
| Model | Model name to use | llama3, codellama |
Ollama Model License Guide
While Ollama itself is MIT-licensed, individual models have their own licenses. When you select a model, ProjectHub automatically classifies its license and displays guidance.
| Classification | Example Models | Display |
|---|---|---|
| Commercial OK | Llama 3, Qwen 2.5, Phi, DeepSeek, Gemma, Mistral | No warning |
| Restricted | Codestral | Red warning |
| Verify Required | Models not in the known list | Orange notice |
When distributing commercially, always verify the license terms of the models you intend to use.
Network Settings (P2P)
Configure P2P connections between ProjectHub instances.
Screenshot: Network settings panel
Network Status Indicator
Screenshot: Network status indicator in the toolbar (connected/disconnected)
| Status | Color | Description |
|---|---|---|
| Connected | Green | Connected to another ProjectHub instance |
| No Connection | Gray | Running in standalone mode |
| Error | Red | Connection failed |
P2P Features
| Feature | Description |
|---|---|
| Project Chat | Real-time chat between collaborators on the same project |
| Contact Sync | User list across the network |
| Authentication | Approve or deny incoming connection requests |
Authentication Dialog
Screenshot: P2P authentication dialog (approve/deny)
When a connection request arrives from another ProjectHub instance, an authentication dialog appears:
- Approve -- Allow the connection
- Deny -- Reject the connection
Security System
ProjectHub uses an HMAC-SHA256 authentication system to secure AI Agent communications.
AI Agent Authentication
| Item | Description |
|---|---|
| Algorithm | HMAC-SHA256 |
| Key Storage | ~/.projecthub/agent_secret.key |
| Token Expiry | 60 seconds |
| Replay Prevention | Nonce-based |
Authentication Flow
1. ProjectHub generates a token using the secret key
- The token is passed to the AI Agent
- The AI Agent authenticates with the token
- The token expires if not used within 60 seconds
- The same token cannot be reused (nonce)
Secret Key Management
The secret key is automatically generated on the first app launch:
- Location:
~/.projecthub/agent_secret.key - Format: Random byte string
- Permissions: Owner read/write only (600)
If you delete the secret key, a new one is automatically generated, but all existing sessions will be invalidated.
App Data Management
Data Locations
| Data | Location |
|---|---|
| Project list | ~/Library/Application Support/ProjectHub/projects.json |
| Paper bookmarks | ~/Library/Application Support/ProjectHub/paper_bookmarks.json |
| Schedule notification settings | ~/Library/Application Support/ProjectHub/ScheduleNotificationSettings |
| Security key | ~/.projecthub/agent_secret.key |
Backing Up Data
To back up ProjectHub's global data:
# Back up global settings
cp -r ~/Library/Application\ Support/ProjectHub/ ~/ProjectHub_Backup/
Back up security key
cp ~/.projecthub/agent_secret.key ~/ProjectHub_Backup/
Resetting Data
Warning: The following operations will delete all project lists and settings.
# Reset global settings
rm -rf ~/Library/Application\ Support/ProjectHub/
Reset security key
rm -rf ~/.projecthub/
Next Steps
- Shortcuts & Tips -- Productivity-boosting shortcuts
- Getting Started -- Start from the beginning
- Project Management -- Complete project management guide