{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreib5hsumjlcyiwjw4b7exjgxwshfpffp4yekg3lykvbbxsew2l22ru",
    "uri": "at://did:plc:pi6woz4d47bkuws673w2il2r/app.bsky.feed.post/3mksjhk46btp2"
  },
  "path": "/t/anti-llm-sentiment-considered-harmful/14008?page=2#post_29",
  "publishedAt": "2026-05-01T14:34:37.000Z",
  "site": "https://discourse.haskell.org",
  "textContent": "shinzui:\n\n> I’m concerned that developers who use coding agents to contribute to the ecosystem may be discouraged from participating.\n\nI think we have to be careful with the wording here.\n\nMaintainers have the autonomy to decide what kind of contributions they accept. That can range from anything involving “I don’t like your commit messages” to “your PR is too big”.\n\nMany of them are only working on these things in their free time and being bombarded with huge slop PRs that you have to look over even more carefully is probably something that not everyone is keen to get into.\n\nYes, there’s more extreme cases, where maintainers categorically decline LLM assisted contributions, but have you considered their reasoning? E.g. for GHCup, I simply think the project is too small to be interesting for LLM assisted contributions and I feel it may compromise the project… because it’s about distribution, where small mistakes can have very large impact.\n\nI think similarly for the core libraries I maintain. These are _foundations_. `unix` is hard to understand and reviewing big PRs in a timely manner is already challenging. Wrong decisions are difficult to revert.\n\nI think these discussions are useful to have and I think it is productive to open issues on the respective projects and ask for a clear LLM policy, if they don’t have one already.\n\n“What is good use of LLMs?” is a difficult question to answer and it’s evolving constantly.",
  "title": "Anti-LLM Sentiment Considered Harmful"
}