Masking Engine

Nothing Sensitive Ever Leaves.

See exactly what PromptWall detects and masks across every category of sensitive data. Real examples. Real patterns. Zero data leaks.

15+
Detection Patterns
Built-in regex patterns covering all major sensitive data types
<5ms
Detection Latency
Regex + policy matching runs in under 5 milliseconds
~400ms
Deep Detection
Optional proprietary contextual detection for edge cases
<2%
False Positive Rate
Precision-tuned patterns minimize false alerts
Zero
Data Stored
Original sensitive values are never stored on our servers
6+
AI Tools Covered
ChatGPT, Claude, Copilot, Cursor, Gemini, Codeium

Try It Yourself

Type or paste any text below and see what PromptWall would detect and mask. This runs entirely in your browser — nothing is sent anywhere.

7 detections:
Credentials (3)Internal Docs (1)Personal Info (3)
Employee types this prompt
Help me debug this API call. Here's my setup:

API Key: sk-proj-abc123def456ghi789jkl012mno345pqr678
Endpoint: https://10.0.4.55:8080/api/v1/users
Auth: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.abc123

The user john.doe@acme.com (phone: +1-555-123-4567)
reported the issue. Their SSN is 123-45-6789.

password=xK9#mP2$vL5nQ8
AI tool receives this
Help me debug this API call. Here's my setup:

API Key: [MASKED_CREDENTIALS_3]
Endpoint: https://[MASKED_INTERNAL_DOCS_1]:8080/api/v1/users
Auth: [MASKED_CREDENTIALS_2]

The user [MASKED_PERSONAL_INFO_3] (phone: [MASKED_PERSONAL_INFO_2])
reported the issue. Their SSN is [MASKED_PERSONAL_INFO_1].

[MASKED_CREDENTIALS_1]

Detected Items

OpenAI Key
sk-proj-abc123def456ghi789jkl012mno34...
Internal IP (10.x)
10.0.4.55
Bearer Token
Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6Ik...
Email
john.doe@acme.com
US Phone
+1-555-123-4567
SSN
123-45-6789
Password
password=xK9#mP2$vL5nQ8

This demo runs entirely in your browser using our Stage 1 regex engine. No data is sent anywhere. The full product adds Stage 2 (custom policies) and Stage 3 (proprietary contextual detection).

What We Detect

15+ built-in patterns across 6 categories, plus custom per-tenant rules and proprietary contextual detection.

Personal Information

john.smith@acme.com
[MASKED_PERSONAL_INFO_1]
+1 (555) 123-4567
[MASKED_PERSONAL_INFO_2]
SSN: 123-45-6789
SSN: [MASKED_PERSONAL_INFO_3]

Financial Data

Card: 4532-1234-5678-9012
Card: [MASKED_FINANCIAL_DATA_1]
Account: 12345678 Routing: 021000021
Account: [MASKED_FINANCIAL_DATA_2] Routing: [MASKED_FINANCIAL_DATA_3]

Health Data

Patient ID: MRN-2024-45892
Patient ID: [MASKED_HEALTH_DATA_1]
HRN: H-789456
HRN: [MASKED_HEALTH_DATA_1]

Credentials & Secrets

sk-proj-abc123def456ghi789jkl012mno345
[MASKED_CREDENTIALS_1]
ghp_1234567890abcdefghijklmnopqrstuvwxyz12
[MASKED_CREDENTIALS_2]
AKIAIOSFODNN7EXAMPLE
[MASKED_CREDENTIALS_3]
Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6...
[MASKED_CREDENTIALS_4]
password=SuperSecret123!
[MASKED_CREDENTIALS_5]

Source Code & Keys

-----BEGIN RSA PRIVATE KEY-----
MIIEpAIBAAKCAQEA...
-----END RSA PRIVATE KEY-----
[MASKED_SOURCE_CODE_1]
postgres://admin:P@ssw0rd@prod-db.acme.com:5432/customers
[MASKED_CREDENTIALS_6]

Internal Infrastructure

Server at 10.0.4.55 port 3306
Server at [MASKED_INTERNAL_DOCS_1] port 3306
VPN: 192.168.1.100
VPN: [MASKED_INTERNAL_DOCS_2]

Real-World Scenarios

These are actual prompts employees send to AI tools every day. See what PromptWall catches — and what the AI tool actually receives.

Developer asking ChatGPT for help

Software Engineer using ChatGPT

4 fields maskedcredentialspersonal infointernal docs
What the employee typed
I'm getting a connection error. Here's my config:

DB_HOST=prod-db-master.acme.internal
DB_USER=admin
DB_PASS=xK9#mP2$vL5nQ8
DB_NAME=customers

The error happens when user john.doe@acme.com tries to log in.
My API key is sk-proj-abc123def456ghi789jkl012mno345pqr678.
What the AI tool receives
I'm getting a connection error. Here's my config:

DB_HOST=[MASKED_INTERNAL_DOCS_1]
DB_USER=admin
DB_PASS=[MASKED_CREDENTIALS_1]
DB_NAME=customers

The error happens when user [MASKED_PERSONAL_INFO_1] tries to log in.
My API key is [MASKED_CREDENTIALS_2].

HR manager using Claude for a letter

HR Manager using Claude

2 fields maskedpersonal info
What the employee typed
Draft an offer letter for:
Name: Priya Sharma
Email: priya.sharma@gmail.com
Phone: +91 98765 43210
Position: Senior Engineer
Salary: 28,00,000 per annum
SSN equivalent: ABCDE1234F (PAN)
Starting: March 15, 2025
What the AI tool receives
Draft an offer letter for:
Name: Priya Sharma
Email: [MASKED_PERSONAL_INFO_1]
Phone: [MASKED_PERSONAL_INFO_2]
Position: Senior Engineer
Salary: 28,00,000 per annum
SSN equivalent: ABCDE1234F (PAN)
Starting: March 15, 2025

DevOps engineer debugging with Cursor

DevOps Engineer using Cursor

3 fields maskedcredentialsinternal docs
What the employee typed
This Terraform isn't working. Here's the relevant config:

provider "aws" {
  access_key = "AKIAIOSFODNN7EXAMPLE"
  secret_key = "wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY"
  region     = "ap-south-1"
}

resource "aws_instance" "prod" {
  ami           = "ami-0c55b159cbfafe1f0"
  instance_type = "t3.large"
  private_ip    = "10.0.1.42"
}
What the AI tool receives
This Terraform isn't working. Here's the relevant config:

provider "aws" {
  access_key = "[MASKED_CREDENTIALS_1]"
  secret_key = "[MASKED_CREDENTIALS_2]"
  region     = "ap-south-1"
}

resource "aws_instance" "prod" {
  ami           = "ami-0c55b159cbfafe1f0"
  instance_type = "t3.large"
  private_ip    = "[MASKED_INTERNAL_DOCS_1]"
}

Finance team querying Gemini

Finance Analyst using Gemini

4 fields maskedpersonal infofinancial data
What the employee typed
Analyze this customer refund:
Customer: Sarah Johnson, sarah.j@bigclient.com
Card ending 4532-8901-2345-6789 was charged $4,500
Account routing: 021000021, Account: 1234567890
Refund reason: duplicate charge on invoice INV-2025-0892
What the AI tool receives
Analyze this customer refund:
Customer: Sarah Johnson, [MASKED_PERSONAL_INFO_1]
Card ending [MASKED_FINANCIAL_DATA_1] was charged $4,500
Account routing: [MASKED_FINANCIAL_DATA_2], Account: [MASKED_FINANCIAL_DATA_3]
Refund reason: duplicate charge on invoice INV-2025-0892

Three Layers of Detection

Every prompt passes through three detection stages. Fast patterns first, then policies, then AI — in under 500ms total.

1

Regex Engine

~1ms

15+ high-precision patterns catch structured data instantly. Emails, phones, SSNs, credit cards, API keys, bearer tokens, private keys, connection strings, internal IPs.

Runs on every request. Zero network calls. No false negatives on known patterns.

2

Policy Engine

~1ms

Custom per-tenant rules with regex patterns and keyword block lists. Block "Project Titan" mentions. Mask internal codenames. Alert on competitor names.

Admins configure rules in the dashboard. Changes sync to agents within 5 minutes.

3

Deep Detection

~400ms

Proprietary contextual analysis catches what regex can't. Business confidential info, narrative PII, schema dumps that imply sensitive data, informal mentions.

Optional (Business plan+). Runs on isolated infrastructure we control. Results merged with regex detections.

Zero Original Data Stored

Masking is one-way. Original sensitive values are replaced in-flight and never stored on our servers.
No token maps or reverse-mapping capability exists. Even we cannot reconstruct the original data.
Audit logs store the masked version only — what was detected, which categories, when. Not the original content.
Non-AI network logs contain metadata only — URL, domain, method, status, timing. No request or response bodies.
GDPR, SOC 2, and DPDP Act compliant by design. Even if our database were breached, there is no sensitive data to leak.
Start Free Trial