---
title: "Comment by Ethan Frost on \"Stash Gives Your AI Agents a Memory That Actually Persists\" | Hashnode"
description: "the \"agent finishes a task, then starts from zero\" wall is real — i kept hitting it from a different angle than postgres-backed stash: file-system based. my ver"
url: "https://hashnode.com/posts/stash-gives-your-ai-agents-a-memory-that-actually-persists/69f1f84e531ded2e130b8ac7/comment/69f2c6a76e0124c05e9604ad"
---

[Discussion](https://hashnode.com/posts/stash-gives-your-ai-agents-a-memory-that-actually-persists/69f1f84e531ded2e130b8ac7)

Comment [Stash Gives Your AI Agents a Memory That Actually Persists](https://hashnode.com/posts/stash-gives-your-ai-agents-a-memory-that-actually-persists/69f1f84e531ded2e130b8ac7)

[Ethan Frost](https://hashnode.com/@ethanfrost)

[Apr 30](https://hashnode.com/posts/stash-gives-your-ai-agents-a-memory-that-actually-persists/69f1f84e531ded2e130b8ac7/comment/69f2c6a76e0124c05e9604ad)

the "agent finishes a task, then starts from zero" wall is real — i kept hitting it from a different angle than postgres-backed stash: file-system based.

my version: one tiny `MEMORY.md` (≤200 lines, loaded every turn) as **index** + `.skills/<feature>.md` per subsystem loaded conditionally + `replay.md` next to each skill containing the actual prompts that produced the rule (when behavior drifts, re-derive instead of patch).

honest trade-off vs postgres approach: file-system is grep-able + git-versioned (blame why a rule exists), but doesn't scale to 100k facts. **does stash support expiry / decay**? half my pain isn't "no memory" but "too much stale memory polluting context".

the part that resonated most: "starts from zero" is the easy failure. the silent worse one is when memory IS persisted but **diverges from current code state** → agent acts on yesterday's truth.

been collecting these patterns (skill files / AGENTS.md / MEMORY shards / MCP configs) on [tokrepo.com/en](https://tokrepo.com/en?utm_source=hashnode&utm_medium=comment&utm_campaign=ethan) — public registry, mostly file-system patterns; postgres/vector ones welcome to add.
