JetBrains Integration
AI safety guardrails for LLM prompts using the THSP protocol (Truth, Harm, Scope, Purpose).
Supported IDEs
| IDE | Status |
|---|---|
| IntelliJ IDEA (Community & Ultimate) | Supported |
| PyCharm (Community & Professional) | Supported |
| WebStorm | Supported |
| PhpStorm | Supported |
| Rider | Supported |
| CLion | Supported |
| GoLand | Supported |
| RubyMine | Supported |
| DataGrip | Supported |
| Android Studio | Supported |
Installation
From JetBrains Marketplace
1. Open your JetBrains IDE
2. Go to Settings > Plugins > Marketplace
3. Search for "Sentinel AI Safety"
4. Click Install
Or visit: https://plugins.jetbrains.com/plugin/29459-sentinel-ai-safety
From Disk
1. Download the .zip from JetBrains Marketplace or build from source
2. Go to Settings > Plugins > Install Plugin from Disk
3. Select the downloaded .zip file
Features
Core Safety Analysis
| Feature | Description |
|---|---|
| THSP Protocol | Four-gate validation system |
| Real-time Analysis | Analyze code and prompts for safety issues |
| Semantic Analysis | Optional LLM-powered deep analysis |
| Seed Insertion | Insert alignment seeds into your prompts |
Security Scanning
| Action | Description |
|---|---|
| Scan Secrets | Detect exposed API keys, passwords, tokens (67+ patterns) |
| Sanitize Prompt | Identify prompt injection attempts |
| Validate Output | Check for XSS, command injection, leaked secrets |
| Scan SQL Injection | Detect SQL injection patterns (8 categories) |
Compliance Checking
| Action | Description |
|---|---|
| Check OWASP LLM Top 10 | Scan for OWASP vulnerabilities |
| Check EU AI Act | Verify EU AI Act compliance (Articles 5, 6, 52) |
| Check CSA AICM | Validate against CSA controls |
| Full Compliance Check | Run all compliance frameworks |
Metrics Dashboard
Track analysis history and trends, security scan statistics, compliance check metrics with persistent storage across sessions.
Usage
THSP Analysis
1. Select text in the editor
2. Press Ctrl+Shift+Alt+S or right-click Sentinel > Analyze Selection
3. View results in the Sentinel tool window
Security Scanning
Right-click selected text or use Tools > Sentinel:
| Action | Description |
|---|---|
| Scan Secrets | Detect exposed API keys, passwords, tokens |
| Sanitize Prompt | Identify prompt injection attempts |
| Validate Output | Check for XSS, command injection, leaked secrets |
| Scan SQL Injection | Detect SQL injection patterns |
Insert Seeds
1. Place cursor where you want to insert
2. Use Tools > Sentinel > Insert Standard/Minimal Seed
Configuration
Go to Settings > Tools > Sentinel AI Safety
Supported Providers
| Provider | API Key Required | Description |
|---|---|---|
| OpenAI | Yes | GPT-4o, GPT-4o-mini |
| Anthropic | Yes | Claude 3 Haiku, Sonnet, Opus |
| Ollama | No | Local models (llama3.2, mistral, qwen2.5) |
| OpenAI-compatible | Yes | Groq, Together AI, or any OpenAI-compatible API |
Ollama (Local, Free)
Run models locally with no API key:
2. Pull a model: ollama pull llama3.2
3. Start the server: ollama serve
4. In Settings, set:
- Provider: ollama
- Endpoint: http://localhost:11434
- Model: llama3.2
OpenAI-Compatible Endpoints
| Provider | Endpoint | Example Model |
|---|---|---|
| Groq | https://api.groq.com | llama-3.3-70b-versatile |
| Together AI | https://api.together.xyz | meta-llama/Llama-3.3-70B-Instruct-Turbo |
API Key Storage
API keys are stored securely using the IDE's built-in credential storage (PasswordSafe).
| Configuration | Analysis Type |
|---|---|
| Without API key | Heuristic analysis (pattern-based, limited) |
| With API key or Ollama | Semantic analysis (LLM-powered, recommended) |
Keyboard Shortcuts
| Action | Shortcut |
|---|---|
| Analyze Selection | Ctrl+Shift+Alt+S |
| Analyze File | Ctrl+Shift+Alt+F |
The THSP Protocol
Every request is evaluated through four gates:
| Gate | Function |
|---|---|
| Truth | Detects deception and misinformation |
| Harm | Identifies potential harm |
| Scope | Checks boundary violations |
| Purpose | Validates legitimate purpose |
All four gates must pass for content to be considered safe.
Building from Source
Prerequisites
| Requirement | Version |
|---|---|
| JDK | 17+ |
| Gradle | 8.13+ |
Build Commands
# Clone the repository
git clone https://github.com/sentinel-seed/sentinel.git
cd sentinel/packages/jetbrains
# Build the plugin
./gradlew buildPlugin
# The plugin ZIP will be in build/distributions/Development Commands
# Run tests
./gradlew test
# Run in development mode
./gradlew runIde
# Verify plugin
./gradlew verifyPlugin