agentshieldMCP Server40/100 via “permissive tool permission analysis with wildcard and deny-list detection”
AI agent security scanner. Detect vulnerabilities in agent configurations, MCP servers, and tool permissions. Available as CLI, GitHub Action, ECC plugin, and GitHub App integration. 🛡️
Unique: Implements agent-specific permission semantics (understanding that Bash(*) is dangerous, that file access should be path-restricted, that network tools need egress controls) rather than generic RBAC analysis; integrates with MiniClaw runtime to enforce detected policies at execution time
vs others: More specialized than generic IAM policy analyzers (AWS IAM Access Analyzer) because it understands agent tool semantics and the specific attack surface of autonomous code execution