Official Model Context Protocol server for the platform, exposing redaction pipelines, policies, and detections as tools for AI agents.
概览
Model Context Protocol server for the Nvisy multimodal redaction platform. Nvisy detects and removes sensitive information across documents, images, and audio. It combines deterministic patterns, NER, computer vision, and LLM-driven classification into auditable, policy-driven pipelines built for regulated industries such as healthcare, legal, government, and financial services. This server gives agent runtimes such as Claude Code and Claude Desktop access to those pipelines. It is built on @nvisy/sdk. Or add it to a client's configuration file: The server is configured through the environment: A typical run is list_files to find a file, list_pipelines to choose a pipeline, redact_file to detect, get_analysis to review, then apply_redaction to produce the redacted output. Files are uploaded through the Nvisy app or API rather than this server, and no tool deletes anything.
README
以下为原文 README 节选，完整内容请查看原文。
@nvisy/mcp
Model Context Protocol server for the Nvisy multimodal redaction platform.
Nvisy detects and removes sensitive information across documents, images, and audio. It combines deterministic patterns, NER, computer vision, and LLM-driven classification into auditable, policy-driven pipelines built for regulated industries such as healthcare, legal, government, and financial services.
This server gives agent runtimes such as Claude Code and Claude Desktop access to those pipelines. It is built on @nvisy/sdk.
Installation
npm install @nvisy/mcp
Quick Start
Run the server over stdio:
NVISY_API_TOKEN=your-api-token npx @nvisy/mcp
Register it with Claude Code:
claude mcp add nvisy --env NVISY_API_TOKEN=your-api-token -- npx -y @nvisy/mcp
Or add it to a client’s configuration file:
{
"mcpServers": {
"nvisy": {
"command": "npx",
"args": ["-y", "@nvisy/mcp"],
"env": {
"NVISY_API_TOKEN": "your-api-token",
"NVISY_WORKSPACE": "your-workspace"
}
}
}
}
The server is configured through the environment:
| Variable | Required | Default | Description |
|---|---|---|---|
NVISY_API_TOKEN |
Yes | — | API token used to authenticate against the Nvisy API |
NVISY_BASE_URL |
No | https://api.nvisy.com |
Base URL of the Nvisy API |
NVISY_WORKSPACE |
No | — | Default workspace slug; every tool can override it |
Tools
| Tool | Description |
|---|---|
list_workspaces |
Workspaces the API token can access |
list_pipelines |
Redaction pipelines in a workspace |
list_policies |
Policies deciding what is detected and how it is redacted |
list_files |
Files stored in a workspace |
list_labels |
Entity labels this deployment can detect |
list_detections |
Detection runs, most recent first |
redact_file |
Run a pipeline over a file and wait for the findings |
get_detection |
Poll a detection’s status |
get_analysis |
Summarise what a detection found |
apply_redaction |
Write a redacted copy of the file |
A typical run is list_files to find a file, list_pipelines to choose a
pipeline, redact_file to detect, get_analysis to review, then
apply_redaction to produce the redacted output.
Files are uploaded through the Nvisy app or API rather than this server, and no tool deletes anything.
Features
- Modern ES2022+ JavaScript target
- Full TypeScript support with strict typing
- Runs standalone over stdio, or embedded as a library
安装
npx -y @nvisy/mcp配置
{
"mcpServers": {
"nvisy": {
"command": "npx",
"args": ["-y", "@nvisy/mcp"],
"env": {
"NVISY_API_TOKEN": "your-api-token",
"NVISY_WORKSPACE": "your-workspace"
}
}
}
}