Repository navigation
Conversation
As LLM use becomes more prevalent, make it clear what expectations are around use of these tools when interacting with this project. This policy is inspired by the policies of several other projects including: * matplotlib * FastAPI * pydantic * Astral (uv/ty/ruff etc) * pandas * SciPy * Rust * LLVM * Podman
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #1777 +/- ##
=======================================
Coverage 96.30% 96.30%
=======================================
Files 50 50
Lines 3821 3821
=======================================
Hits 3680 3680
Misses 141 141 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
| the LLM, time should be spent increasing understanding around the project before | ||
| attempting a fix. | ||
|
|
||
| LLM use MUST be disclosed where used. |
There was a problem hiding this comment.
Perhaps "LLM generated code must be disclosed"
abbiemery
left a comment
There was a problem hiding this comment.
All things I agree with. Just a comment on being explicit with our expectations
| the LLM, time should be spent increasing understanding around the project before | ||
| attempting a fix. | ||
|
|
||
| LLM use MUST be disclosed where used. |
There was a problem hiding this comment.
We should be explicit here about what is meant by use and how this ought to be disclosed. Perhaps if any code was generated then you must say "This code was partially generated by claude" in the PR description and final commit message.
There was a problem hiding this comment.
I normally add this:
AI Disclaimer
The code and this description were generated with AI and reviewed/tweaked by the author.
| LLM output MUST NOT be used in any communication to other maintainers. This | ||
| includes: | ||
|
|
||
| * Issues |
There was a problem hiding this comment.
I disagree with this one, I quite like @danielballan's and @hyperrealist issue and PR template
My bug report
I've got this bug, it's important to me because of blah
-- Me, a human person
AI generated details
Wall of text with things that are useful to everyone, things that are useful for someone, and things that are useful for no-one in it| * Issues | ||
| * PR descriptions | ||
| * PR reviews | ||
| * Commit messages |
There was a problem hiding this comment.
Also disagree on this one, the robot writes much better commit messages than me
| * PR descriptions | ||
| * PR reviews | ||
| * Commit messages | ||
| * Comments/replies |
There was a problem hiding this comment.
absolutely agree with this one though, LLMs should be driven by humans, and not allowed to talk to anyone else
| the LLM, time should be spent increasing understanding around the project before | ||
| attempting a fix. | ||
|
|
||
| LLM use MUST be disclosed where used. |
There was a problem hiding this comment.
I normally add this:
AI Disclaimer
The code and this description were generated with AI and reviewed/tweaked by the author.
| Documentation is intended for people to read and it is strongly recommended that | ||
| it is written manually. |
There was a problem hiding this comment.
Agree for docs. Disagree for ADRs. If you sketch a design and get an LLM to fill it out, and it finds problems with your sketch, it is in a good position to write the decision record about how you got to the final design. This is something that would otherwise be lost, and is useful when you read the code later and want to know why the code looks the way it does
As LLM use becomes more prevalent, make it clear what expectations are
around use of these tools when interacting with this project.
This policy is inspired by the policies of several other projects
including: