Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Log is the Agent

arxiv.org

Post reply on HN
Print view View original

11–20 of 59 posts

Re: The Log is the Agent

#11

Post by spike021 »

jamiegregz wrote: post #9

> In this arrangement the log is a byproduct: an audit artifact written alongside the real computation, never the substrate of it. I’ve come to the same conclusion building my own agents. It simply feels ‘wrong’ that most frameworks will happily mutate your context. You have to explicitly go out of your way to store the original events. I’ve now started storing an event log for my own agents, this is used as the sour…

Why not save progress and important results of a conversation (i.e. including tool calls and such) to a project markdown (even multiple as needed) and clear your context window completely rather than compacting many times? You can then just specify a markdown file to be included as context. Especially if following any kind of plan document and executing on a part of it.

Re: The Log is the Agent

#13

Post by jamiegregz »

spike021 wrote: post #11
jamiegregz wrote: post #9

> In this arrangement the log is a byproduct: an audit artifact written alongside the real computation, never the substrate of it. I’ve come to the same conclusion building my own agents. It simply feels ‘wrong’ that most frameworks will happily mutate your context. You have to explicitly go out of your way to store the original events. I’ve now started storing an event log for my own agents, this is used as the sour…

Why not save progress and important results of a conversation (i.e. including tool calls and such) to a project markdown (even multiple as needed) and clear your context window completely rather than compacting many times? You can then just specify a markdown file to be included as context. Especially if following any kind of plan document and executing on a part of it.

[dead]

Re: The Log is the Agent

#19

Post by tern »

lukebuehler wrote: post #5

Very cool. I settled on the same/similar design in my agent harness. All relevant events that affect the context window are stored in an event log. Forking agents and sessions is simply setting a pointer to the sequence number of another event log. So if you want to check an implementation of this pattern see: https://github.com/smartcomputer-ai/lightspeed

Nice work. Excited to check this out.

Re: The Log is the Agent

#20

Post by try-working »

lmwnshn wrote: post #8
bigcat12345678 wrote: post #3

This is true after learning this framing. It's more like the log is the only user/agent accepted consensus. It has to be the grounding base. Although extending it into an agentic system architecture becomes something not necessarily effective in practice.

With my database hat on, in the context of agentic systems I would argue that write-ahead logs form a good (and potentially transactional) interface between speculative agent work and durable world mutations [0]. That said, there are a _lot_ of "logs for agents" papers that I've read (and unfortunately gotten assigned to review) which are basically "we asked claude to hack on a graph DB and generate a paper". [0] htt…

We should probably only interact with the agent by writing to the log, which it executes from, and the agent should probably only interact with the external environment by writing and executing code. That fixes a lot of issues with non-determinism.
Post reply on HN