alt Hacker NewsAgents don't need memory, they need documentation
Comments
We’ve been working on exactly this, deriving documentation from external systems (like GitHub, etc) into a giant set of docs that agents can reference and edit. Works way better than I would’ve originally expected. Suddenly these things are able to reference granola notes, a slack discussion, and an RFC when making coding decisions. Super helpful in a lot of unexpected ways!
A refreshing perspective on agentic architecture. Shifting the focus from bloated contextual memory to structured, verifiable documentation and state boundaries is precisely the right systems-level trade-off.
I never understood memory solutions for coding agents.
You have the whole session history right there. One recall skill and some JSON parsing gets you grep over perfect memory. Why would you ever use more tools to spend more tokens to construct a imperfect memory next to your session history?
I just don't get it.
Peter Naur argued in his famous essay Programming as Theory Building that documentation alone cannot fully capture or preserve the complete mental model behind a program.
However, AI works differently from humans in that much more of its working context has to be made explicit. Because of that, there may be some fundamentally different way for AI to maintain or reconstruct a program’s overall model.
Or you can just tell agent to grep your past sessions.
I don't use memory. I am really not keen on the idea of having this hidden context fed into the LLM, and I prefer to use docs, both markdown and stored in a wiki like Confluence. Using repos and wikis also lets you scope the information so hyper-specific memory doesn't affect other projects.
If I want the LLM to remember something I ask it to update some docs, and even check that docs are consistent across the board after doing so.
The only thing I want the LLM to remember everywhere is talk like a human (no load-bearing, not this/that etc...), so I have an AGENT.md for that.
I think many AI fanatics will focus on a wrong point. They will say "just use AI to generate the fucking documentation. What are you talking about?"
Indeed, AI fanatics lack critical thinking. They easily accept the idea that AI needs documentation, but merely disagree the method to approach it.
Yes, this makes sense. Memory is an uncurated and often opaque system of arbitrary past discussions. It can help, it can harm. Accurate documentation in the other hand is only beneficial.
Agents need BOTH documentation and multiple "memory" systems that also point to your docs and source. There are a lot of mature options out there, but post and the proposed solution both fall short.
All of this is nonsense. The right way to think of LLMs correctly is that they are a one stop shop for all the questions in the universe that you speak to in simple natural language. Don’t overcomplicated things and use esoteric magic spells like we’ve had in tech for decades. You end up getting the best results this way because you’re not jamming the context full of detailed minutiae.
on my claude.md I said don't write any memory bitch no one need that fucking dogshit features
[flagged]
Obligatory 927.
[flagged]
[dead]
I’ve noticed a growing pattern of people creating repositories full of Markdown documentation, or adding large amounts of it directly to their main repositories, often generated by agents. In some cases, this can add up to megabytes of material, and I haven’t yet seen much evidence that this level of documentation meaningfully improves an agent’s performance and that it just doesn't rot over time.
If someone has good A/B evals of this being more effective I'll eat my hat, but the reason no one is publishing them is because well, evals are hard, and this is likely just magical thinking.
My current view is that an agent generally needs three things to work effectively: a way to discover information that isn’t obvious, such as a minimal `AGENTS.md` that points to more focused brief files; a clear way to verify that its work is correct; and some guidance on project-specific tastes. Everything else is noise.
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[dead]
[flagged]
[flagged]
[flagged]
[dead]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
[flagged]
What's with this vector database obsession?