3 ms·
IMO, these agentic guard rails aren't the answer. It seems like we're seeing that the more you stuff context, the more the agents forget and don't follow the gu
by cautiouscat 2mo ago
IMO, these agentic guard rails aren't the answer. It seems like we're seeing that the more you stuff context, the more the agents forget and don't follow the guidelines.[1]
Things like ArchUnit, static analyzers, and other deterministic tools can help with lower level things like architecture. For higher up stuff, I am increasingly feeling like agents don't guarantee anything and in many cases its just the opposite. This is where a thoughtful engineer and reviewer can keep things in check.
It's possible I'm off base here, but I can't make heads or tails of the LLM written readme.
1: https://arxiv.org/html/2510.05381v1 https://arxiv.org/html/2510.05381v1
- kanfilior 2mo agosee the architect adr and Product pdr, also the whole idea is to have an index of context directive similar to skills which the modal invoke
- taude 2mo agoOur skills and agents use deterministic tools for things like: linting, ayy1, etc....
- enraged_camel 2mo agoYeah. One of the most productive things we did was run a dynamic workflow with Fable to scan for issues in the repo, then have it go through the findings and determine which ones could have been found with deterministic tooling, either tooling that already exists, or could be created from scratch and wired up. The resulting changes we made have been paying off in droves.
- jondwillis 2mo agoI’ve found hooks to be really useful for correcting or inducing certain behaviors/outcomes that tend to occur during agentic development. For example, I’m working on a project to remake the Final Fantasy XI client. My repository has a bunch of git submodules that reference other peoples’ related efforts, and an open source server. For example, despite including CLAUDE.md to suggest otherwise, Claude Code always ends up writing these insanely dense comments referring to specific files and lines of code in submodules. When those submodules update, now the comments are no longer correct. So I added a hook to detect when comments are included. Then, for example, I have a deterministic heuristic and script involved to remove some kinds of comments, and another that asks Haiku to quickly LLM-as-a-judge whether or not to edit/remove the comment. Similarly, I have a stop hook that reminds Claude to commit its code logically on main, noting that other changes may have been added by other concurrent sessions (I avoid worktrees and even branching for this particular project and stage of development.) It works well.