AI-powered psychological profiling tool that analyzes video content to generate comprehensive behavioral assessments using multimodal analysis.
Created by @LLMSherpa
Upload a video file (or paste a YouTube/Twitter/TikTok URL) and the system runs a 6-stage analysis pipeline:
| Step | Name | Description |
|---|---|---|
| 1 | Video | Load, validate, and prepare video for processing |
| 2 | Audio | Extract audio track and transcribe speech |
| 3 | Visual | Analyze facial expressions, body language, behavioral archetypes |
| 4 | Multi | Cross-reference visual and audio cues, detect performance vs authenticity |
| 5 | Voice | Examine vocal patterns, accent, paralinguistic markers |
| 6 | Synth | Combine all analyses into final FBI-style behavioral profile |
Upload a video or import from URL. The 6-step progress bar tracks analysis in real-time.
Expandable sections for each analysis type: Behavioral Archetype, Multimodal Analysis, Audio/Voice, LIWC Linguistics, and Speech Transcript.
Interactive charts including:
- Confidence Gauge - Overall analysis confidence
- Big Five Radar - OCEAN personality dimensions
- Dark Triad Bars - Narcissism, Machiavellianism, Psychopathy scores
- Threat Matrix - Risk assessment levels
- MBTI Profile - Cognitive function preferences
Chase Hughes / NCI University methodology for behavioral deception indicators:
- BTE Score - Behavioral Table of Elements deception probability
- Blink Rate Analysis - Stress indicator tracking
- FATE Model - Focus, Authority, Tribe, Emotion motivational drivers
- Integrated Deception Assessment with conflict resolution
- Python 3.10+ (3.11 recommended)
- FFmpeg (required for audio extraction)
- OpenRouter API key - Get one here
git clone https://github.com/shitcoinsherpa/The-Profiler.git
cd The-ProfilerThe setup script will automatically download FFmpeg and install all dependencies.
Windows:
setup.batLinux/macOS:
chmod +x setup.sh run.sh
./setup.shWindows:
run.batLinux/macOS:
./run.shThe app will open in your browser at http://localhost:7860
- In the web UI, expand Settings & Configuration
- Paste your OpenRouter API key
- Click Save API Key
- Click Test Connection to verify
- Upload a video file (10-300 seconds, max 100MB) or paste a URL
- Optionally enter a subject name to track profiles over time
- Click Initiate Behavioral Analysis
- Watch the 6-step progress bar (~1-3 minutes depending on video length)
- Review results in the three output tabs:
- Analysis Components - Individual analysis breakdowns
- FBI Profile & Insights - Synthesized profile with visualizations
- NCI Deception Analysis - Chase Hughes methodology indicators
- Export as PDF or JSON
| Tab | Contents |
|---|---|
| Analysis Components | Behavioral Archetype, Multimodal Analysis, Audio/Voice, LIWC Linguistics, Transcript |
| FBI Profile & Insights | Personality charts (Big Five, Dark Triad, MBTI, Threat Matrix), FBI Synthesis, Confidence scores |
| NCI Deception Analysis | BTE Score gauge, Blink Rate chart, FATE Model radar, Deception Conflict Matrix |
Below the main interface, the Profile History section lets you:
- Browse previously analyzed subjects
- Compare multiple profiles for the same person
- Review past analyses without re-running
In the Settings & Configuration accordion:
| Setting | Description |
|---|---|
| API Key | Your OpenRouter API key (encrypted storage) |
| Vision Model | Model for visual analysis (default: Gemini 2.5 Pro) |
| Multimodal Model | Model for audio+visual analysis (default: Gemini 2.5 Flash) |
| Synthesis Model | Model for final profile synthesis (default: Gemini 2.5 Pro) |
| Max Resolution | Video processing resolution (720p/1080p) |
| Frame Interval | Seconds between frame samples |
| Cache | View/clear cached results |
In the Custom Prompt Templates accordion, you can:
- Create custom analysis prompts
- Save templates for different use cases
- Modify the default FBI synthesis prompt
- Native Video Processing - Sends video directly to Gemini's multimodal understanding
- Parallel Analysis Pipeline - Multiple sub-analyses run concurrently for speed
- Result Caching - Avoids redundant API calls for the same video
- Profile Database - SQLite storage with search and history
- PDF Reports - Professional formatted export
- NCI Methodology - Chase Hughes behavioral science framework
- Deception Conflict Matrix - Reconciles contradicting signals across modalities
| Issue | Solution |
|---|---|
| "FFmpeg not found" | Run setup.bat again, or manually download from gyan.dev and place ffmpeg.exe in the project folder |
| "API key invalid" | Verify your key at openrouter.ai/keys and re-enter |
| "Video too long/large" | Videos must be 10-300 seconds, under 100MB |
| Analysis fails midway | Check OpenRouter credits; some models have usage limits |
| Charts not showing | Ensure plotly installed: pip install plotly |
MIT
For educational and research purposes only. AI-generated psychological profiles are speculative assessments, not clinical diagnoses. Do not use for employment, legal, or medical decisions.


