Overcoming this kind of nondeterministic behavior around creating/following/modifying instructions is the biggest thing I wish I could solve with my LLM workflows. It seems like you might be able to do this through a system of Claude Code hooks, but I've struggled with finding a good UX for maintaining a growing and ever-changing collection of hooks.

Are there any tools or harnesses that attempt to address this and allow you to "force" inject dynamic rules as context?

Rather, I think the point being made was that we need to reframe the problem at hand; instead of trying to "control misinformation" maybe we need to work on ways to _amplify_ true information in ways that render misinformation powerless.

This sentiment is dying at an alarming rate. Even here, after reading some of these comments I am taken aback at how normalized the idea of total surveillance has become. It doesn't matter where it comes from, big corp or big gov, taking away personal privacy takes away a fundamental part of what makes us human. It's not healthy. It's not right.

Fight this.