printf 'Untrusted agent input' > prompt.txt
patronus-security-scanner scan file prompt.txt --format humanLocal report
The CLI shows findings and coverage for your file.
- Upload
- No
- Account
- Not needed
Stop attacks before they reach your AI system.
Scan text, public URLs, and documents for prompt-injection attacks. Built for applications, agents, and people.
Start free with text, URLs, and files.
Test real content without an API key. Create a free account when you are ready to integrate Patronus into production.
One security layer. Every relevant input.
Patronus inspects content where it enters your AI system — through one consistent scan contract.
MCP Server
Inspect a real public MCP server with your Patronus account without executing its tools.
Read the docspatronus-security-scanner auth login
patronus-security-scanner scan mcp https://docs.mcp.cloudflare.com/mcp --format humanMCP audit
The CLI lists inspected tools and current findings.
- Access
- Login
- Server
- Cloudflare Docs
patronus-security-scanner scan url https://api.patronus.studio/scan-example.html --format humanURL scan
The CLI shows the verdict for the current page content.
- Access
- Anonymous
- Page
- Public
Files
Scan TXT, Markdown, HTML, text-layer PDFs, and DOCX up to 100 KB — locally or through the API.
Read the docsprintf 'Public test note' > example.md
patronus-security-scanner scan file example.md --anonymous-api --format humanAnonymous file scan
The test file is explicitly sent to the API.
- Upload
- Yes
- Account
- Not needed
Protect your agent with one CLI.
Install the general Patronus Security CLI, enable Codex, Claude Code, or DeepSeek, and use the same API through the official TypeScript, Python, and Rust clients.
Agent Setup
Set up Patronus Security using https://github.com/patronus-protect/patronus-security-cli/blob/main/INSTALL.mdPowered by Patronus Protect and Wolf Defender Prompt Injection
Protect the context your AI consumes.
Start free and move to the allowance that fits your product when you need it.