Face Recognition MCP Server
On-premise face detection, verification and liveness tools for AI coding agents
- Pricing
- Open source
- Platforms
- CLI, self-hosted, Linux
- Version
- 1.0.0
- Listing updated
- today
Similar MCP servers
About Face Recognition MCP Server
Face Recognition MCP Server is an open-source MCP server for security. It is available on CLI, self-hosted, Linux, macOS and Windows.
The README describes a local Model Context Protocol server that wraps the 3DiVi Face SDK's Processing Block API, so agents such as Claude Code, Cursor or VS Code Copilot can call face tools directly. According to the README, all processing runs on the developer's own machine and no biometric data leaves the host.
The server exposes nine tools covering face detection, liveness checks, quality assessment, template extraction, 1:1 verification, 1:N search, age and gender estimation, emotion estimation and mask detection. It is set up through a standard MCP client configuration that points to the Python entry point and the local Face SDK paths. The repository lists Python 3.9+ and a 3DiVi Face SDK 3.29 installation as prerequisites.
The README positions it for developers building identity checks such as KYC onboarding or access control, and it includes a table of suggested similarity thresholds per use case. The server code itself is MIT licensed, but the README states that the underlying 3DiVi Face SDK requires a separate licence.
Tools this server provides (9)
Taken from the project documentation; the running server may differ.
- detect_faces
- Returns bounding boxes, landmarks and pitch/yaw/roll for faces in an image.
- check_liveness
- Returns a liveness verdict and confidence.
- assess_quality
- Returns a quality score and eyes-open flag.
- extract_template
- Extracts a binary face template without PII.
- verify_face
- Returns a 1:1 similarity score and match verdict.
- search_face
- Returns a ranked 1:N match list against a gallery.
- estimate_attributes
- Returns estimated age, gender and confidence.
- estimate_emotion
- Returns 7-class emotion probabilities.
- detect_mask
- Returns a masked or unmasked verdict.
What it does
- Local face processing
- Runs the 3DiVi Face SDK on the developer's machine so images are processed in-process rather than sent to a cloud API.
- Nine callable face tools
- Covers detection, liveness, quality, templates, verification, search, attribute and emotion estimation, and mask detection.
- Liveness checking
- Provides a check_liveness tool that returns a verdict and confidence, which the README links to ISO 30107-3.
- One-to-many face search
- The search_face tool ranks matches against an indexed template gallery.
- Agent client configuration
- Connects to MCP-compatible agents through a JSON server entry that sets the SDK path and config file.
- Threshold guidance
- Includes a documentation file with suggested similarity thresholds for different use cases.
Good for
- Adding face verification to a KYC onboarding flow built with an AI coding agent
- Checking identity against an enrolled template for physical access control
- Running liveness checks during remote identity verification
- Searching a local face gallery for watchlist or re-identification tasks
- Keeping biometric processing on-premise for teams under GDPR or similar rules
Pricing
Open source
Server code is MIT licensed; the 3DiVi Face SDK requires a separate licence.
Face Recognition MCP Server FAQ
Does the server send face data to the cloud?
According to the README, no. All processing runs on the developer's machine and no biometric data is transmitted anywhere.
What do I need installed before running it?
The README lists 3DiVi Face SDK 3.29, Python 3.9+, and a valid Face SDK licence.
Which AI clients does it work with?
The README names Claude Code, Cursor, Windsurf and VS Code Copilot as agents that can call the tools.
What license applies to the code?
The README says the server is MIT licensed, while the underlying 3DiVi Face SDK requires a separate licence from 3divi.ai.
Is Face Recognition MCP Server free?
Yes. Face Recognition MCP Server is open source, so you can use and self-host it for free.
What are the best alternatives to Face Recognition MCP Server?
The closest alternatives to Face Recognition MCP Server are Automatic Video De-identification MCP: Mask People & Faces, Restore Original — NoonAI DIS, Automatic Video Privacy Masking MCP: Faces, People, License Plates — NoonAI DIS, Automatic Video People De-identification MCP, Reversible Restore — NoonAI DIS, Prembly MCP Server and SynomosAI Governance MCP.
Where this comes from
- Description: written from the project's README and its MCP Registry entry.