[rrvsh]

maintaining an LLM wiki has been a lot more effort than I thought, at least if we are trying to maintain a high quality in structure and writing comprehension (for easier lookups both for the agent and human). Are people just shotgunning their agent wikis or how

[bad_username]

> maintaining an LLM wiki has been a lot more effort than I thought

Same here. Wikis start out good, but either devolve in a journal-y mess after a while and many updates, or require constant expensive rewrites. (I didn't use the software of the OP.)

[reacharavindh]

Wikis start out good, but get stale too quick and become useless or worse confusing.

I’ve experienced this over and over again to strongly believe it.

I genuinely wonder if throwing LLMs at this problem would solve it at least to some extent. Make a LLM agent whose sole purpose is to act as a librarian. It periodically reviews _all_ of the wiki and validates them against codebases, newer docs, anything. Whatever it finds, it should be allowed to intelligently quiz the team/dev whether something is right/stale/wrong and updated it accordingly.

If one tolerates that toil - answering questions of a library bot, would it result in a usable wiki base?

[rrvsh]

I think writing the instructions for such an agent would be almost as much work as maintaining the wiki and learning your preferences yourself (and both may be coupled intrinsically unless you already manage a large enough notes system). The crucial problem is that the agents lack 1. taste and 2. the ability to know what is good for them. I've bounced off trying these like three times over the last couple of years, twice after Feb 2026, simply because it requires way too much toil that would be better (IMO) put into maintaining your own knowledge base, where at least that toil would result in learning

[DenisM]

In short, yes. The effort will amortize for a larger team over longer time period.

https://openai.com/index/harness-engineering/