Security MCP servers for scanning, threat intelligence and audits: let an AI assistant check code, dependencies and infrastructure for risks. 3,540 servers in this category.
Provides a two-tier autonomous reverse engineering framework that combines fast CLI reconnaissance with deep formal verification and CPU emulation.
Enables AI agents to access logged-in browser sessions without exposing sensitive credentials to the model or chat logs.
Provides a governance boundary for AI agents by enforcing risk-tiered execution, human approval, and audit trails for API calls.
Provides a security runtime for autonomous coding agents to enforce authority boundaries, verify evidence, and manage human approval gates.
Provides open-source execution authority for AI agents through consequence preview, trust routing, and signed cryptographic receipts.
Scans Solana token mints to detect cabal activity, rug risks, funding clusters, and honeypots before an AI agent executes a trade.
Provides Claude with specialized tools for GCC business compliance, security auditing, HR/payroll calculations, and safe read-only database access.
Acts as an OAuth 2.1 proxy that injects API keys or custom headers into requests, allowing OAuth-only clients to connect to header-based MCP servers.
Acts as a runtime security guard for MCP servers by requiring human approval for write operations and detecting tainted tool outputs.
Scans MCP client configuration files for leaked credentials, hardcoded secrets, and unsafe settings before an agent process runs.
Creates tamper-evident decision logs by storing salted SHA-256 digests in an evidence log while keeping full records in a local journal.
Enforces human-in-the-loop approval for smart home device actions through a secure proposal and execution workflow.
Enables AI agents to securely exchange files through scanned, provenance-stamped, and expiring capability URLs with human approval gates.
A vulnerable-by-design MCP server and Semgrep rule set for practicing and detecting object-level authorization bugs like BOLA and IDOR.
Provides a secure environment for AI agents to use secrets by name without ever seeing their actual values.
Manages a structured person roster and permission system to control what AI agents can do with specific users.
Scans dependency manifests against OSV.dev to identify known vulnerabilities and malicious packages in npm, PyPI, Go, Rust, and more.
Lets an AI client validate phone numbers, identify carriers, detect disposable numbers, and assess risk scores without making a call.
Convert MCP server configurations between Claude, VS Code, Cursor, and Windsurf while linting for embedded secrets.
Provides a security layer for AI agents by requiring human passkey approval for irreversible API actions and shell commands.
