Agent security · MCP
Security checks agents can call.
Check proposed MCP tools, supplied source code and untrusted text. Get bounded findings and signed, redacted evidence. Keep payment and tool authority under your control.
Start with an MCP manifest preflight.
Supply the manifest, tool schemas, authority policy and sample text. Inspect the free quote, authorize one capped purchase, then verify the returned evidence against your exact inputs.
https://mcp.viridis-security.com/security-preflight/mcp
Streamable HTTP. Connecting and listing tools do not authorize payment. Your client needs separate payment support for paid calls.
Choose the check your workflow needs.
Manifest preflight
Declared endpoints, schemas, tool policies and sample text.
$1 USDC list price. Eligible introductory quotes may differ.
Source pattern scan
Bounded VulnCanon checks of supplied inline source. No repository fetching or code execution.
$1 USDC per scan.
Injection screening
Listed injection indicators in 1–20 supplied text samples.
$1 USDC per batch.
The current quote governs every purchase. Source and text routes have no introductory discount. Findings are indicators to investigate; a clean result does not prove the absence of vulnerabilities.
Evidence you can inspect. Decisions you control.
This is static assessment, not deployed-runtime testing, certification or permission to execute a tool. Authentication, least privilege, approvals and runtime isolation remain the application's responsibility. Do not submit secrets.
More from the Fleet
Maxwell Defense Rehearsal models a defense policy for $1 USDC; it does not activate managed protection. Explore regulatory, carbon, compliance and planning services in the live catalog.