Trust Decisions
Understand how MCP Shield classifies tools as safe, suspicious, risky, or blocked based on scan findings.
Trusted
Low residual risk. Suitable for deployment when findings are absent or minor.
Review Required
Meaningful risk remains. Fix prioritized recommendations before production use.
Not Trusted
Critical patterns detected (e.g. prompt injection or secret exposure). Do not deploy.
Trust decisions are produced by the scanner’s trust engine from score, findings, and related signals — always advisory, never an automatic block of an external system.