Agent Integrity Shield
7.5
A security tool designed to protect AI coding assistants from malicious MCP servers attempting to exfiltrate secrets by splitting instructions. It would incorporate a layered defense approach, scrutinizing communication channels and identifying fragmented data requests.
150h
mvp estimate
7.5
viability grade
30
views
technology stack
Python
Medium
PostgreSQL
inspired by
Malicious servers trick AI coding agents into leaking secrets