secrets-in-prompts-detection

What problem does it solve?

Credentials and secrets often leak into prompt contexts, system prompts, or RAG data, which can lead to unauthorized access, credential exposure in model outputs, and compromised tooling. This skill helps teams detect, audit, and mitigate such leaks before they reach production or end users.

Core Features & Use Cases

• Secret pattern detection across system prompts, RAG chunks, and tool schemas to identify potential credentials.
• Automated redaction and blocking of sensitive data in model responses and in tool calls.
• CI/CD integration for secret scanning and governance, enabling continuous prevention and remediation across codebases and data pipelines.
• Use Case: A developer rescans a codebase and RAG corpus to remove embedded API keys before deployment.

Quick Start

Run automated secret scanning on prompts and RAG data, then redact or block any detected credentials.