A CLI tool that scans any public MCP server for SSRF, missing auth, and stdio RCE flaws before a developer adds it to their agent config
Between January and April 2026, 40+ CVEs were filed against MCP servers. BlueRock Security scanned 7,000+ public MCP servers and found 36.7% have SSRF vulnerabilities and 41% require no authentication at all. OX Security disclosed a systemic RCE in the MCP SDK's stdio transport affecting 150M+ downloads, and Anthropic's own mcp-server-git shipped with three RCE-enabling flaws that were quietly patched. Every developer adding a third-party MCP server to their agent config is implicitly trusting code that, statistically, has a one-in-three chance of SSRF exposure and nearly even odds of requiring no auth at all. This tool lets a developer run a single command against any public MCP server's GitHub repo URL or running endpoint and get a line-level trust report covering SSRF patterns in tool parameter handling, auth posture on tool routes, prompt injection strings in tool descriptions, and unsanitized shell calls in stdio transport handlers, before the server ever touches their agent config.
Demand Breakdown
Social Proof 2 sources
Gap Assessment
3 tools exist (mcp-scan (Invariant Labs), Pillar Security, Trend Micro AI Security) but gaps remain: Does not scan source code for SSRF or stdio RCE patterns, does not check auth posture on endpoints, runs post-install not pre-adoption, and has no CVE feed integration for known-bad servers; No MCP-server-specific static analysis, no pre-adoption CLI vetting workflow, targets enterprise deployments not individual developers adding community MCP servers.
Features2 agent-ready prompts
Competitive LandscapeFREE
| Product | Does | Missing |
|---|---|---|
| mcp-scan (Invariant Labs) | Open-source scanner that checks MCP server tool definitions for prompt injection and tool-poisoning patterns by reading the tool descriptions returned at runtime | Does not scan source code for SSRF or stdio RCE patterns, does not check auth posture on endpoints, runs post-install not pre-adoption, and has no CVE feed integration for known-bad servers |
| Pillar Security | Enterprise AI security platform covering LLM app scanning and agent threat detection at the network layer | No MCP-server-specific static analysis, no pre-adoption CLI vetting workflow, targets enterprise deployments not individual developers adding community MCP servers |
| Trend Micro AI Security | Research and enterprise threat intelligence covering AI agent attack surfaces; published findings on 492 zero-auth exposed MCP servers | No developer-facing CLI or scanning tool, research output only, no actionable pre-adoption gate for individual developers |
Leads24BUILDER
Sign in to unlock full access.