AgentShield Inspector Pro
MCP inventory security analysis, policy validation, baseline comparison and reports.
How to pay
Subscribe
$9.99/month
Predictable monthly cost with included usage. Best for steady, high-volume traffic.
- Unlimited tools within plan limits
- One API key, billed once a month
- Cancel any time
AgentShield MCP Inspector Pro helps developers and security teams review MCP tool capabilities before integrating them into AI workflows.
Analyze supplied MCP inventories with 20 security rules and 6 capability-combination checks. Identify signals involving command execution, file access, credentials, network requests, database operations, messaging, financial actions, privilege escalation, and other sensitive capabilities.
Each finding includes a risk level, supporting metadata evidence, and recommended remediation.
PRO FEATURES
• Configurable tool allowlists and blocklists. • Trusted and blocked domains, plus filesystem path policies. • Custom rule severity and documented exceptions with optional expiration. • Detection of dangerous capability combinations within individual tools and across a server. • Baseline comparisons highlighting added or removed tools, schema changes, new capability signals, and increased risk. • JSON, Markdown, and SARIF reporting.
AVAILABLE MCP TOOLS
• inspect_inventory — Analyze supplied tool definitions and optional policy settings. • compare_scans — Compare two AgentShield JSON scan reports. • validate_policy — Check policy configuration for errors. • list_security_rules — Explore supported rules and capability combinations.
HOW IT WORKS
Provide an MCP inventory containing tool names, descriptions, and input schemas. AgentShield evaluates the supplied metadata and returns findings classified as SAFE, REVIEW, HIGH RISK, or CRITICAL.
The hosted tools analyze supplied data only. They do not execute inspected tools, launch target servers, fetch external URLs, or read your local files. Hosted scans support up to 100 tools per inventory, subject to request-size and processing limits.
SCOPE
AgentShield provides metadata-based risk analysis, not a runtime security audit or safety certification. Findings indicate advertised capabilities and possible combinations—not proven malicious behavior. A SAFE result means no configured risk signals were detected; it does not guarantee that a server is safe.
Built for MCP integration reviews, policy checks, and security review of changing tool inventories.