AgentSentinel
A transparent MCP proxy that inspects every tool response for prompt injection attacks before the content reaches the model's context window.
AgentSentinel wraps any MCP server as a transparent proxy, scanning tool responses in real-time for prompt injection patterns before they reach the model. In detect mode it logs threats to a structured NDJSON feed; in block mode HIGH-severity injections are dropped before the model sees them. Ships with 8 detection patterns across LOW/MEDIUM/HIGH severity tiers and a live threat dashboard on GitHub Pages.
The build log
Follow the work.
New essays and prototypes, in your feed reader. Email subscriptions are paused while I connect a working delivery service. No email is collected here.
Follow via RSS