How do I use Presidio MCP Server?

1. Click on "Install Server". 2. Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state. 3. In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Presidio MCP Server analyze PII in 'John Doe, SSN 123-45-6789'" That's it! The server will respond to your query, and you can continue using it as needed. Here is a step-by-step guide with screenshots.

Presidio MCP Server

by ManoharMarri

Overview Schema Related Servers Score Discussions

Python

Remote

Presidio MCP Server

A production-ready Model Context Protocol (MCP) server that exposes Microsoft Presidio PII detection and anonymization capabilities using FastMCP 2.x.

Features

Primitive	Count	Description
🔧 Tools	8	PII analysis, anonymization, batch processing, risk scoring
📦 Resources	4	Entity catalogue, operator guide, server config, examples
💬 Prompts	4	Audit reports, sharing workflows, HR reviews, batch checks

Tools

Tool	Description
`analyze_text`	Detect PII entities with confidence scores & character offsets
`anonymize_text`	Anonymize PII with configurable per-entity operators
`deanonymize_text`	Reverse AES-encrypted anonymization
`batch_analyze`	Analyze up to 50 texts in one call
`list_supported_entities`	Return all detectable entity types
`add_custom_recognizer`	Dynamically register regex-based recognizers at runtime
`score_pii_risk`	Compute a 0–100 PII risk score with recommendations
`list_recognizers`	List all active recognizers with metadata

Resources (read-only reference data)

URI	Description
`presidio://entities/catalogue`	All supported PII types grouped by category
`presidio://operators/guide`	Anonymization operator reference with examples
`presidio://config/server`	Live server configuration snapshot
`presidio://examples/common`	Ready-to-use usage examples

Prompts (reusable workflow templates)

Prompt	Use case
`pii_audit_report`	Generate a compliance-ready PII audit report
`anonymize_for_sharing`	Anonymize a document before external sharing
`hr_data_privacy_review`	HR-specific PII review and anonymization
`batch_pii_policy_check`	Run policy compliance checks across record batches

HR-Domain Custom Recognizers

Beyond Presidio's built-in NLP, this server includes custom recognizers for:

EMPLOYEE_ID — EMP-XXXXXX patterns
SALARY — Currency amounts in USD/GBP/EUR + k-notation
UK_NINO — UK National Insurance Numbers
US_SSN — US Social Security Numbers (improved patterns)
BANK_ACCOUNT — UK sort codes + IBAN prefixes
PASSPORT — US and UK passport number formats

Project Structure

mcp-server/
├── presidio_mcp/
│   ├── __init__.py          # Package metadata
│   ├── __main__.py          # Module entry point (python -m presidio_mcp)
│   ├── server.py            # FastMCP server — tools, resources, prompts
│   ├── engines.py           # Singleton Presidio engine initialization
│   ├── models.py            # Pydantic input/output models
│   └── recognizers.py      # HR-domain custom recognizers
├── tests/
│   └── test_presidio_mcp.py # Comprehensive test suite
├── pyproject.toml           # Project config & dependencies
├── requirements.txt         # pip-installable dependencies
└── README.md

Setup & Installation

1. Prerequisites

Python 3.10+
pip or uv (recommended)

2. Create virtual environment

cd mcp-server
python3 -m venv .venv
source .venv/bin/activate     # Windows: .venv\Scripts\activate

3. Install dependencies

# Production
pip install -r requirements.txt

# Or with dev tools (testing, linting):
pip install -e ".[dev]"

4. Download the spaCy NLP model

# Recommended (higher accuracy, ~741MB):
python -m spacy download en_core_web_lg

# Lightweight fallback (~12MB, auto-used if lg is missing):
python -m spacy download en_core_web_sm

Running the Server

stdio transport (Claude Desktop, Cursor, Continue)

python -m presidio_mcp
# or
presidio-mcp

HTTP/SSE transport (remote clients, testing)

python -m presidio_mcp --transport sse --host 0.0.0.0 --port 8001

Claude Desktop Integration

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "presidio": {
      "command": "/path/to/mcp-server/.venv/bin/python",
      "args": ["-m", "presidio_mcp"],
      "cwd": "/path/to/mcp-server"
    }
  }
}

Running Tests

pytest tests/ -v

Test coverage includes:

All 8 tools (happy paths + edge cases)
All 4 resources (JSON validity, content checks)
All 4 prompts (message structure, keyword presence)
Validation errors (blank text, boundary scores)
Multi-language support

Anonymization Operators Quick Reference

Operator	Reversible	Best For
`replace`	✗	General de-identification
`redact`	✗	Complete removal
`mask`	✗	Partial masking (credit cards)
`hash`	✗	Consistent pseudonymization
`encrypt`	✅	Reversible by authorised parties

Example Usage (via LLM)

Analyze HR document:

Use analyze_text with:
  text: "New hire John Smith (EMP-001234), SSN 123-45-6789, salary $95,000"
  entities: ["PERSON", "EMPLOYEE_ID", "US_SSN", "SALARY"]

Anonymize for sharing:

Use anonymize_text with:
  text: "Contact Alice at alice@corp.com, card: 4111-1111-1111-1111"
  operators:
    - entity_type: EMAIL_ADDRESS, operator: replace
    - entity_type: CREDIT_CARD, operator: mask, chars_to_mask: 12

Risk-score a document:

Use score_pii_risk with:
  text: "Full employee record with SSN, bank account, and salary details"

Architecture Decisions

Decision	Rationale
Singleton engines	Presidio engine init is expensive (~2-5s). One instance per process.
Pydantic models as tool params	FastMCP uses model schemas for LLM tool descriptions + strict validation
Graceful spaCy fallback	Auto-falls back to `en_core_web_sm` if `en_core_web_lg` is missing
`readOnlyHint` annotations	Read-only tools skip confirmation prompts in MCP clients
Thread-safe double-checked locking	Engines are safe for concurrent asyncio tool calls
stderr for logs	MCP stdio protocol uses stdout; logs go to stderr to avoid interference

This server cannot be installed

license - not tested

quality - not tested

maintenance

How are these scores calculated?

Maintenance

–Maintainers

–Response time

–Release cycle

–Releases (12mo)

Commit activity

Resources

GitHub Repository

Need Help?

Related Servers

Unclaimed servers have limited discoverability.

Looking for Admin?

If you are the server author, to access and configure the admin panel.

Latest Blog Posts

Who's Calling? MCP Hosts Are an Identity Blind Spot (And the Spec Knows It)
By Om-Shree-0709 on July 25, 2026.
mcp
Agent Identity
OAuth 2.1
Your AI Chatbot Just Exposed Your CEO's Salary to an Intern
By Om-Shree-0709 on July 2, 2026.
Agent Identity
MCP Security
OAuth Delegation
Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)
By Om-Shree-0709 on June 30, 2026.
Agentic Ai
Prompt Injection
WebAssembly

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/ManoharMarri/local-presidio-mcp-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server