Looking for feedback on the framework. Cross agent validation
retained text
I'm building a open source framework for coding agents. The main thing it does is catch ai slop before it reaches the repo. In the demo, the same prompt goes to the same model twice. Without the framework, the agent finished, the tests passed, and a bug shipped. With it, a second agent reviewed every change automatically and sent its findings back. It also has a session manager, a ledger and guards for mistakes a model keeps repeating. The video goes through those too. Video in the first comment. Happy to answer any questions.
document fetched 2026-10-31 00:18 · canonical https://reddit.com/r/AI_Agents/comments/1xjnzb0/looking_for_feedback_on_the_framework_cross_agent