Products
The ZeroHarmAI open-source security toolkit helps teams detect sensitive data, secure agent workflows, and build stronger controls into AI delivery.
zero-harm-ai-detectors
Enterprise-grade AI safety tooling that protects applications from exposing PII, intellectual property, and other sensitive content.
Core Features
- 🛡️Sensitive data detection and redaction
- 🔌Easy Python library integration
- 🧑💻Developer Friendly API
- 🤖AI-Powered Detection or Fast Heuristic Detection
- 🤖ML trainable for custom enterprise needs
Try It Live
Test our PII or harmful content detection right in your browser. Try entering an email, phone number, SSN or harmful text.
zero-harm-ai-gha
zero-harm-ai-gha is a GitHub Action and CLI tool for detecting AI-specific security risks in pull requests before merge. It scans changed code for prompt injection risk, secrets exposure, PII leaks, unsafe LLM tool usage, prompt logging, and sensitive data flowing into LLMs.
Core Features
- 🛡️AI security scanning for pull request diffs
- 💬GitHub Actions annotations and reviewer-style feedback
- 📄JSON, Markdown, and SARIF outputs
- 🔎Detection powered in part by zero-harm-ai-detectors
What It Detects
- •Prompt injection risk
- •PII and secret exposure
- •Sensitive data flowing into LLMs
- •Unsafe LLM tool or function usage
- •Prompt and message logging
- •System prompts containing secrets
zero-harm-ai-mcp
zero-harm-ai-mcp is an MCP server for AI safety and privacy workflows. It gives applications and agents access to detection and redaction tooling for sensitive, risky, and harmful content through MCP-native integrations.
Core Features
- 🧩MCP-native access to AI safety tooling
- 🔒Detection and redaction for sensitive and risky content
- 🤖Designed for agent, automation, and developer workflows
- ⚙️Built to integrate with broader Zero Harm AI safety tooling
Use Cases
- •Agent safety guardrails
- •PII and secret screening
- •Prompt and content redaction workflows
- •MCP-powered application integrations