Is the future of defense-against-foreign-agents-on-my-codebase to subtly hide prompt injections into one’s codebase that would defeat agents to find security bugs ?

If the attackers of ffmpeg need to be using such those authors’ services to find RCE in popular tools to attack, what the ffmpeg team needs to defeat attackers is to reduce efficiency of such tools depthfirst

No...