How can we automate the disclosure of what AI agent was used in a PR and the extent of code? Would be nice to also have an audit of prompts used, as that could

How can we automate the disclosure of what AI agent was used in a PR and the extent of code? Would be nice to also have an audit of prompts used, as that could also be considered “code”.