Developer creates AEP (Agent Experience Protocol) to let AI remember user preferences and workflows across conversations, eliminating repetitive constraint setup.

AEP saves successful AI workflows including intent, constraints, preferences, steps, failure traps, and success checks as repo-local JSON files in .agent/aep/

Works agent-agnostic and functions similarly to Git for AI interactions rather than traditional prompts or memory systems

Addresses the core problem that AI conversations restart from zero each time, forcing users to re-explain preferences and workflows repeatedly

Creator exploring integration potential with tools like Cursor, ChatGPT, and Claude to standardize workflow reuse across AI platforms