Skillware skills · security/deepfake_guard
Deepfake & Document Authenticity Guard
Air-gapped media & document authenticity guard for AI agents — multi-signal forensic verification for synthetic media, digital tampering, C2PA provenance credentials, and ICAO 9303 MRZ passport forgery with zero external API calls.
Capabilities
- ICAO 9303 (7, 3, 1) cyclic check digit validation on TD1, TD2, TD3 passports & IDs
- C2PA JUMBF manifest, digital watermark, and generative AI metadata extraction
- Error Level Analysis (ELA) JPEG recompression differential analysis
- High-pass Laplacian noise residual variance & copy-move spatial cloning detection
- 2D FFT moiré peak ratio analysis for screen recapture photo detection
- ISO/IEC 7810 ID-1 card geometry and aspect ratio verification
- Strictly air-gapped local execution — zero outbound network requests or SaaS leaks
Problems this skill solves
People search for tasks — not always "skillware":
- detect deepfake image ai agent
- offline document authenticity verification python
- validate passport mrz icao 9303 python
- check c2pa provenance credentials agent
- error level analysis ela python agent
- ai agent kyc document forgery detection
- screen recapture moire detection agent
- verify identity document authenticity air gapped
- synthetic media detection python tool
- laplacian noise residual variance image forensics
- agent preflight kyc document validation
Install
pip install skillware then SkillLoader.load_skill("security/deepfake_guard")
FAQ
- How does an AI agent verify passport and ID authenticity offline?
- Use Skillware security/deepfake_guard — pip install "skillware[security_deepfake_guard]", load the skill, and run validate_mrz or inspect_document. It computes ICAO 9303 (7, 3, 1) check digits across TD1, TD2, and TD3 zones with zero network calls.
- Can Skillware detect synthetic media or AI-generated images locally?
- Yes — deepfake_guard combines radial spectral slope roll-off, C2PA JUMBF provenance metadata, JPEG Error Level Analysis (ELA), and Laplacian noise residuals strictly on-device without cloud API dependencies.
- How does deepfake_guard catch fraudsters photographing computer screens in KYC?
- It runs 2D Fast Fourier Transform (FFT) high-frequency peak ratio analysis to detect regular LCD/OLED raster grids (moiré patterns) present when a screen is recaptured by a camera.