nshkrdotcom/GUARDRAIL

Implements a layered security architecture for MCP-based LLM applications with four core components: an Information Gateway Layer (filtering and sanitizing LLM outputs), Message Classification System (threat detection), Trust Scoring Engine (attestation between services), and Flow Control mechanisms (resource limits and rate-limiting). Built as a protocol-agnostic framework with practical, incremental adoption in mind—developers can implement individual security layers rather than requiring full-framework deployment.

No commits in the last 6 months.