agentseal and agentshield
These are **competitors** offering overlapping AI agent security scanning capabilities—both detect vulnerabilities in agent configurations and MCP tool permissions, though AgentSeal emphasizes supply chain attacks and prompt injection testing while AgentShield focuses on broader configuration auditing across multiple deployment formats (CLI, GitHub Action, etc.).
About agentseal
AgentSeal/agentseal
Security toolkit for AI agents. Scan your machine for dangerous skills and MCP configs, monitor for supply chain attacks, test prompt injection resistance, and audit live MCP servers for tool poisoning.
Runs a six-stage detection pipeline combining pattern matching, deobfuscation, semantic embeddings (MiniLM-L6-v2), and baseline SHA-256 hashing to identify malicious configs across 28+ supported agents. Includes 225+ adversarial probes for prompt injection testing—using deterministic canary string detection rather than LLM judges—plus real-time file watching via `shield` and live MCP server auditing that scans tool descriptions for hidden instructions before they reach the agent.
About agentshield
affaan-m/agentshield
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. 🛡️
Performs static analysis on Claude Code configs with 102 security rules across secrets, permissions, hooks, and MCP servers, automatically distinguishing between active runtime exposure and project templates via `runtimeConfidence` scoring. Includes an optional three-agent Opus 4.6 adversarial analysis mode for deeper vulnerability discovery, plus auto-fix capabilities for safe issues like hardcoded secret replacement with environment variable references.
Scores updated daily from GitHub, PyPI, and npm data. How scores work