Find your next MCP server
Find remote and local MCP servers by task, transport, capability, access, and publisher, with clear verification labels.
14,189 MCP servers
Vendor-operated and community listings, with live-check and source-verification states shown independently.
| Server | Type | Category | Access | Setup |
|---|---|---|---|---|
GuardBee MCP Config AuditorScans Cursor, Claude, Windsurf, and VS Code MCP configurations for unpinned versions, secrets, and typosquats. | CommunityPackage | Developer tools | Source reviewed | |
GuardBee MCP DB GatewayA community MCP server that acts as a gateway to databases, enabling agents to query data while masking personally identifiable information (PII) and secrets. | CommunityPackage | Developer tools | Source reviewed | |
GuardBee MCP Elicitation AuditorAn MCP server package that scans server code for elicitation anti-patterns defined in the MCP specification. | CommunityPackage | Developer tools | Source reviewed | |
GuardBee MCP LLM RedteamAn MCP server package that actively red-teams live LLM endpoints and chatbots. | CommunityPackage | AI and machine learning | Source reviewed | |
@guardbee/mcp-oauth-auditor MCPScans an MCP server's own authorization code for OAuth 2.1 anti-patterns named in the MCP spec's Security Considerations — token passthrough, missing audience validation,… | CommunityPackage | Developer tools | Source reviewed | |
GuardBee MCP Rug Pull DetectorAn MCP server that connects to another MCP server via stdio or HTTP, baselines its tools/list response, and detects unauthorized changes to tool descriptions, schemas, or… | CommunityPackage | Developer tools | Source reviewed | |
GuardBee MCP Security ProxyAn open-source MCP gateway that sits between agents and tools to mask PII, block prompt injection, and maintain audit logs. | CommunityPackage | Developer tools | Source reviewed | |
@guardbee/mcp-server-auditor MCPScans other MCP servers' tool definitions for insecure patterns including excessive agency, shell/eval/SQL/SSRF sinks, loose schemas, hardcoded secrets, and wildcard CORS. | CommunityPackage | Developer tools | Source reviewed | |
@guardbee/mcp-tool-poisoning-scannerA security scanner for Model Context Protocol (MCP) servers that detects tool poisoning attacks and confused-deputy vulnerabilities. | CommunityPackage | Developer tools | Source reviewed | |
GuardBee Memory Poisoning Scanner MCPScans agent code for memory poisoning, detecting untrusted input written into persistent, cross-session memory such as MemGPT-style archival or long-term vector stores that is… | CommunityPackage | AI and machine learning | Source reviewed | |
GuardBee Model Scanner MCPScans ML model files including PyTorch, pickle, safetensors, Keras/H5, and ONNX formats for supply-chain risks such as dangerous pickle deserialization globals, disguised or… | CommunityPackage | AI and machine learning | Source reviewed | |
GuardBee Prompt Injection Scanner MCPA community MCP server that scans RAG content and scraped pages for indirect prompt injection attacks. | CommunityPackage | AI and machine learning | Source reviewed |
Browse by category
Let your agent find its tools
Add Koskamo to your agent to find official MCP servers for a task, inspect access requirements, and get setup links. Public discovery needs no key.
How listings and verification work
Remote listings are live-checked and re-checked weekly. Local packages are validated from their attributed registry or catalog source and are never executed by Koskamo. Community and vendor evidence are labelled separately.
Access labels describe remote connection and discovery. Individual tools may require accounts, credentials, permissions, or payment. Registry membership and endpoint checks are not security audits.
List your own server