agent-chaperone
github.com/agent-chaperone/agent-chaperoneAbout
agent-chaperone is a screening layer that inspects an AI agent's tool calls before they run and their results before the agent reads them, blocking risky, off-task, or secret-leaking actions as well as prompt injections hidden in tool output. It works as an MCP proxy combined with a hooks adapter for a client's built-in tools.
Why it made the leaderboard
Adds a defensive screening layer against prompt injection and risky tool calls for agent harnesses that otherwise execute tool results blindly, addressing a real agent-security gap.
Tags
ai-agentssecurityprompt-injectionmcp-proxy
Tech Stack
Node.jsTypeScript
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.