待翻译：Mycelium AI Policy

AI 服务暂时不可用，以下为来源正文，待恢复后补全翻译。

Uh oh! There was an error while loading. Please reload this page. Notifications You must be signed in to change notification settings Fork 35 Star 779 File tree Expand file treeCollapse file tree Open diff view settings Filter options CONTRIBUTING.md Expand file treeCollapse file tree Open diff view settings Display the source diff Display the rich diff Original file line numberDiff line numberDiff line change @@ -4,6 +4,81 @@ thanks for your interest in contributing to Mycelium, an alleged "operating 44 system"! please bear in mind that this is a hobby project that i work on in my 55 spare time, so pull requests may not always be reviewed in a timely manner. 66 7+ ## AI policy 8+ 9+ when contributing to Mycelium, an alleged "operating system", you are free to 10+ use any tools you deem appropriate. this includes the LLM of your choosing, 11+ agents, loops, gastown, AI code review tools, Amazon Rufus, or even vim. 12+ however, all contributions to this project (including my own) must abide by 13+ exactly one rule: 14+ 15+ your code (and any associated prose, such as commit messages and 16+ documentation) must not _look_ like it was written by a large language model. 17+ 18+ basically, the game is simple: contributors are free to use whatever AI tools 19+ they wish when contributing to mycelium and associated projects, but in order to 20+ have your PRs merged, you have to be able to *trick me* into believing it was 21+ written by a human being. any PRs which appear to have been generated by a model 22+ will be closed as a violation of the Mycelium AI Policy (even if they were 23+ actually written on pencil and paper by a human being in a cave with no access 24+ to the internet). repeated violations of the Mycelium AI Policy may result in a 25+ ban from contributing to the repo. 26+ 27+ ### advice on AI policy compliance 28+ 29+ + when using an AI coding agent, the easiest way to ensure your contributions 30+ are compliant with the Mycelium AI Policy is to read the code that the model 31+ generates, and rewrite any parts of it that "feels like Claude-babble". repeat 32+ this process as needed until it no longer feels like it was written by a 33+ model. 34+ + you may feel that it is possible to prompt an LLM into outputting code that 35+ doesn't look model-generated. my personal experience has been that this is 36+ impossible, and model-written code always requires some degree of rewriting to 37+ make it look Normal. if you think you're better at prompting than me, you are 38+ welcome to try it, but...good luck! 39+ + while contributors are free to disclose the use of AI tools in their pull 40+ requests, note that this will put you at a substantial disadvantage, because 41+ an AI disclosure will mean I go in suspicious. if you can trick me into 42+ believing that a pull request which says "This code was written using Claude" 43+ was written entirely by an organic, free-range human being...well, I'll be 44+ impressed. 45+ 46+ ### rationale 47+ 48+ the Mycelium AI Policy may, at first blush, seem a bit strange: why am I 49+ explicitly *encouraging* potential contributors to attempt to mislead a 50+ maintainer? and, doesn't the "you have to trick me" policy seem a bit... 51+ adversarial? to answer some of these questions, allow me to share some of the 52+ thinking behind the Mycelium AI Policy, and why I believe it is the best way to 53+ achieve *my personal goals* for this project. 54+ 55+ 1. this project is a labor of love. I am doing this purely for fun, in my spare 56+ time. reviewing thousands of lines of slop is not particularly fun. reading 57+ the bizarre blend of Rationalist slang, technical jargon, and nonsensical 58+ metaphors that Claude's comments always seem to regress into is particularly 59+ infuriating and I don't ever want to see it in my weekend project. 60+ 3. I believe that code can be beautiful. I love code which is both technically 61+ effective *and* a form of creative self-expression. I want the code in 62+ Mycelium to be beautiful. and I have never found purely model-generated code 63+ to be beautiful. 64+ 4. however, I have also found LLMs to be an effective tool for generating large 65+ amounts of code faster than a human being can type it. i have also found them 66+ to be useful tools in applications where the model is *not* writing the code: 67+ while an LLM review may not question the high-level design of a feature the 68+ way a human reviewer would, they *are* remarkably thorough, and good at 69+ finding the one place where you forgot to change something or didn't use 70+ checked arithmetic or whatever. i don't want to completely forbid the use of 71+ a tool which i *have* found useful in some particular, restricted ways. 72+ 5. i don't want to merge code that is not well-understood by a human being. i do 73+ not believe that it is possible for an engineer to take responsibility for 74+ code that they do not understand thoroughly. in LLM-assisted programming, 75+ there is an obvious temptation to not read the code that the model generates. 76+ we can all fall victim to this temptation from time to time. in my own work, 77+ i have found that the act of rewriting the model-generated code to not Look 78+ Like That forces the person who generated it to actually read and understand 79+ it thoroughly. 80+ 6. it's my project and I can do whatever the hell I want to. 81+ 782 ## building & running 883 984 mycelium is built using [inoculate], the horrible mycelium build tool. @@ -147,4 +222,4 @@ config file, i'd welcome a PR to add support for your preferred package manager. 147222 [direnv]: https://direnv.net/ 148223 [cargo nextest]: https://nextest.sh 149224 [justfile]: #justfile 150- [Just]: https://just.systems 225+ [Just]: https://just.systems 0 (0)