{"id":"agentseal-awesome-mcp-security","name":"awesome-mcp-security","af_score":23.8,"security_score":0.0,"reliability_score":5.0,"what_it_does":"awesome-mcp-security is a curated registry/listing of MCP servers along with a daily-updated “Trust Score” intended to help AI agents assess the safety of using particular MCP servers (i.e., attack-surface/prompt-injection/tool-poisoning risk), not the underlying code quality of those servers.","best_when":"You need a practical starting point to prioritize which MCP servers to integrate, and you will still enforce agent safety controls (sandbox, allowlists, strict permissions).","avoid_when":"You require authoritative or auditable security assurance for a specific MCP server release; this repo provides aggregated scores/curation rather than verifiable per-deployment security guarantees.","last_evaluated":"2026-03-30T15:34:43.142033+00:00","has_mcp":false,"has_api":false,"auth_methods":[],"has_free_tier":false,"known_gotchas":["Scores are intended as agent attack-surface guidance, not a guarantee of safety or correctness of the MCP server itself.","A daily-updated registry can lag behind new MCP server changes; reassess for the exact version/commit you deploy.","Trust scores can be impacted by how tools are used in an agent’s prompt/planning loop; runtime controls remain necessary."],"error_quality":0.0}