Best MCP servers for security
12 working options, checked Nov 8, 2026.
Short answer: the top picks for security right now are InjectScan, LeakGuard and PkgGuard. We list 12 options here: 10 work without sign-in, 1 need a sign-in and 1 run on your own machine.
Security MCP servers protect AI agents and the systems around them. They scan for prompt injection, find leaked secrets, check packages for known problems and test whether other MCP servers can be trusted.
Top picks
-
InjectScan
Scan text, web pages and MCP tool definitions for prompt injection before your agent reads them.
Tools include
scan_text,scan_url,scan_tools -
LeakGuard
Find and redact leaked secrets, wallet keys and PII before an AI agent sends text.
Tools include
scan_for_leaks -
PkgGuard
Check dependencies and lockfiles for known vulnerabilities and malicious packages before your agent installs them.
Tools include
scan_dependencies,check_package -
Wardzo
Scan and monitor websites for leaked keys, exposed files, security headers, SEO issues and uptime.
Tools include
scan_site,get_report,list_sites,add_site,get_site_status -
VerifyMCP
Look up independent trust scores, tool surfaces and change history for MCP servers.
Tools include
list_servers,get_server,get_server_comparison,list_products,list_known_malware
All 12 compared
| Server | What it does | Setup | Tools |
|---|---|---|---|
| InjectScan | Scan text, web pages and MCP tool definitions for prompt injection before your agent reads them. | Hosted, no sign-in | 3 |
| LeakGuard | Find and redact leaked secrets, wallet keys and PII before an AI agent sends text. | Hosted, no sign-in | 1 |
| PkgGuard | Check dependencies and lockfiles for known vulnerabilities and malicious packages before your agent installs them. | Hosted, no sign-in | 2 |
| Wardzo | Scan and monitor websites for leaked keys, exposed files, security headers, SEO issues and uptime. | Hosted, no sign-in | 12 |
| VerifyMCP | Look up independent trust scores, tool surfaces and change history for MCP servers. | Hosted, no sign-in | 9 |
| MCP Checkup | Signed health checks for public MCP servers | Hosted, no sign-in | 6 |
| TestedMCP | We install and test MCP servers: handshake, tool count, context cost and deprecation status for 187 servers. | Hosted, no sign-in | 4 |
| mcp-liveness | Checks whether an MCP server actually answers for a stock client, with a verdict and a one-line fix. | Hosted, no sign-in | 5 |
| agentcheck | Drift and regression monitoring for production AI agents and MCP servers. | Hosted, no sign-in | 11 |
| ActSeal | Ed25519-signed receipts for agent actions — public verify, free 100/day. | Runs locally | – |
| Surface | Preflight a URL before your agent fetches it: unwraps short links (t.co, bit.ly, lnkd.in) and returns fetch or skip. | Hosted, no sign-in | 1 |
| Skycloak MCP | Managed Keycloak identity MCP for AI agents | Hosted, sign-in | – |
How to choose
- Protect the inputs. If your agent reads web pages or email, scan that text for prompt injection.
- Protect the outputs. Redact secrets and personal data before an agent sends text elsewhere.
- Check what you install. Scan dependencies and look up a server's history before adding it.
How to connect one
Here is InjectScan as an example. In Claude Code, run:
claude mcp add --transport http injectscan https://api.youraitools.co/injectscan/mcpFor other apps, follow our guides for Claude, ChatGPT, Cursor and VS Code.
How we picked these
We read each listing's tools and put the closest fit for this job first. Hosted servers must pass a real MCP handshake in our daily check to stay on this page, and local servers are reviewed from their source or package. That shows a server responds, not how well it works for you, and we don't run every tool in production.
Frequently asked questions
Which of these MCP servers work without signing in?
InjectScan, LeakGuard, PkgGuard, Wardzo and VerifyMCP are among the 10 that answered our latest live check with no account or key.
What is prompt injection in MCP?
Prompt injection is when text an assistant reads, such as a web page or tool result, contains hidden instructions that try to change what it does.
How can I check if an MCP server is safe?
Read its source if available, check who runs it, and limit its permissions. Trust-score servers add independent signals, but none can prove a server is safe. See our safety guide.
How often is this list updated?
Every hosted server here is re-checked daily with a real MCP handshake, and a server drops off this page while it fails that check. Last check: Nov 8, 2026.