This email requests a vote on the AGENTS.md PR here: https://github.com/apache/opennlp/pull/1470
This is the initial draft. It is a file for LLMs that all the frontier apps read. It does not promise that LLMs will adhere to the AGENTS.md, but they "usually" do. Per Martin's advice, I suggest extending this to everyone: avoid overthinking or over-engineering. The Linux kernel is dropping the rule to include "assisted by" for a multitude of reasons. Also, keep in mind that it's concise because we want to avoid sending too many tokens. Below is for reference in case you're not familiar with AGENTS.md; we already have a discussion thread, but feel free to discuss here or in the PR itself. I'll summarize everything on this thread as we go along. Official Spec https://agents.md/ Claude's take If CLAUDE.md doesn't exist, it will treat AGENTS.md the same way: https://code.claude.com/docs/en/memory OpenAI's take https://learn.chatgpt.com/docs/agent-configuration/agents-md Lucene's take We did take inspiration from lucene's AI Policy but theirs is for humans, and AGENTS.md is for machine: apache/lucene@c94ed04 <https://github.com/apache/lucene/commit/c94ed045b8a107aef4c2c07eafad42e7b82dbda6> *Linux Foundation's recent take* https://www.phoronix.com/news/Linux-Dropping-Assisted-By-Req Before commenting, I ask that we read some of the above guidelines to understand what the file should and should not contain. Keep in mind, I suspect this will be a moving target - so having something is better than nothing. Again - this file is for machines to read. Not humans. So that's why it's slightly chunked, follows guidelines, and includes nearly everything we discussed on the discussion thread. Kristian
