andreirx

Yes I think this is the current "standard" - keeping notes in tools like Obsidian and having agents query it, or peppering README.md files throughout the codebase. I call it "breadcrumbs" because I just leave things in the path for the agent to pick up on. But it doesn't answer questions like what are all the callers of this function? where is this data going? where is it coming from? So I'm trying to solve THIS part which is much more time consuming and prone to drifting in time.

Better vibe coding - maybe. I wanted to distance myself from vibe coding - trying to get agents to output more deterministic code as opposed to vibing. (if I understand vibe coding correctly it's stating your intention and letting yourself go wherever the agents are taking you)

I also think agentic coding is not going anywhere so I'm looking for ways to make it more appropriate for my job.

Hi HN, I wrote this after spending the last two years on the AI coding adoption curve.

Most takes I'm reading are about treating AI failures as a memory or context window problem. I argue it's an orientation problem. LLMs are probabilistic-like generation engines; enterprise codebases are deterministic structures. Forcing probability into a deterministic system without strict boundaries generates compounding divergence between the end goals and what AI agents are coding.

I've been experimenting with building a deterministic orientation substrate (repo-graph) using a three-layer truth model to force agents to respect boundaries before they generate syntax.

I also argue that traditional safety-critical processes are about to become highly relevant to standard agentic development, not only for compliance, but because heavy process is a sound containment vessel for AI-generated entropy.

Curious to hear how other teams are preventing architectural drift when deploying agents at scale.