Checks mapped to OWASP MCP Top 10

Review your MCP
setup risks.

ShieldMCP checks your configuration for permission risks, possible secrets, and supply chain warning signs. Static analysis is a starting point, not proof that your setup is safe.

Review configuration flags before connecting agents to sensitive systems.

Your config is sent to our server for static analysis. Use a sanitized copy under 50 KB: replace credentials with placeholders before uploading. This can affect secret-detection findings. Results are retained for seven days from scanning or starting checkout. Anyone with the scan URL can access the results, including paid details after unlock. Keep it private. Privacy and data retention

Don't know where your config is? View guide

How it works

Three steps to a safer MCP setup.

01

Upload config

Drop your file or paste JSON. Supports claude_desktop_config.json, .cursor/mcp.json, and more.

02

Static analysis

Configuration checks mapped to OWASP MCP Top 10 categories. No account needed.

03

Fix issues

Review flagged issues. Unlock remediation guidance and config examples where available; test changes in your environment.

What we check

Static checks mapped to OWASP MCP Top 10 categories; runtime behavior is not tested.

ID
Category
Description
MCP01
Token Mismanagement & Secret Exposure
Secrets & API keys in plaintext
MCP02
Privilege Escalation via Scope Creep
Over-broad filesystem, database & API scope
MCP03
Tool Poisoning
Malicious or unverified MCP packages with hidden directives
MCP04
Software Supply Chain Attacks & Dependency Tampering
Unverified packages, unpinned versions & dependency confusion
MCP05
Command Injection & Execution
Shell execution & dangerous commands
MCP06
Intent Flow Subversion
External content ingestion as indirect injection vector
MCP07
Insufficient Authentication & Authorization
Insecure connections, missing auth & public binds
MCP08
Lack of Audit and Telemetry
Missing logs for agent actions
MCP09
Shadow MCP Servers
Hidden or unverified server endpoints
MCP10
Context Injection & Over-sharing
PII over-sharing & source+sink exfiltration paths

Simple pricing

Start free. Unlock details when you need them.

Free

$0

Always free

  • Risk score
  • Category flags
  • Issue titles
  • Server ratings
Run Free Scan
FULL REPORT

Full Report

$49one-time

Per scan report

  • Everything free +
  • Remediation guidance
  • Examples where available
  • Priority order
  • Shareable PDF

Pro

Coming soon
Coming soon

Planned features — not currently available. Pricing to be announced.

  • Planned API access (CI/CD)
  • Planned rescan alerts
  • Planned scan history
  • Planned team configs
  • Planned Slack alerts

What a scan can tell you

Configuration flags help prioritize manual review. The scanner does not connect to your MCP servers, enumerate live tools, or verify effective permissions. A high score or a category with no flags does not establish safety. Review package sources and runtime behavior separately.