# Kevin Park, Founding Engineer (Solo) at Stackporch — read of Agent Memory, June 18 2026

> 7 years backend at mid-size SaaS, went independent 8 months ago, building AI tooling in between client contracts. Kids wake me at 6:15. I listen to podcasts on the 20-minute drive back from school drop-off.

## How I got here

Heard "agent memory" mentioned on a Changelog episode in passing and searched the phrase on Google. This page came up second or third. I was specifically looking for anything that hooks into VS Code and persists context between Claude sessions because I keep losing architectural decisions when my context window resets. I clicked expecting a product, not a pitch deck shop.

## What I clicked first

"Never lose design decisions to context limits" landed. That is the exact sentence I would have typed into a search bar. So I kept reading.

Then I scrolled and hit "Add to VS Code" and clicked it. Nothing happened, or nothing I could see. I clicked it a second time. I'm still not sure what it was supposed to do.

## Where I paused

The scoring table. "$-3,788 Year-1 take-home (Fermi)" and "1 in 6 Meaningful-success odds" printed right there on the page, in the open. That stopped me. Most idea marketplaces project hockey sticks. These people are projecting a loss. I read that section twice. Either this is a genuinely unusual level of honesty or it's a reverse-psychology positioning play. I don't know which. But I didn't leave.

## What I distrusted

I wanted to buy a VS Code extension that stores agent memory. What this page is actually selling is a $99 "strategy package" for someone who wants to BUILD that extension. That mismatch is significant and the page doesn't flag it in the hero at all. I had to read halfway down before I understood the actual product.

Also: "landing page quality: 2/10" is listed as a concern on the page about the landing page I am currently reading. I appreciate the self-awareness. I am also living inside the proof of that score right now.

"We shipped the strategy package; you ship the customer conversations." That line is doing a lot of work. It's honest. It's also the kind of sentence that makes me wonder if anyone has actually built this from one of these packages and talked to a real customer.

## What would convince me

One person who unlocked the $5 dossier, built the thing, and has 20 paying users. Not a testimonial quote. A Loom video, a GitHub repo link, an indie hackers post. Anything that proves the dossier produces a working product and not just a well-organized Notion doc.

Also: show me what the VS Code integration actually does in a GIF or a two-minute video. "Add to VS Code" as an unresponsive button is not a demo.

## What I'd ask in an email reply

1. Has anyone who bought the $99 adopt package shipped a product from it? Can I talk to them?
2. What is the actual mechanism the MVP uses to store memory across context windows? Is it a local file, a vector store, a sidecar process? The "design decisions" framing implies structured storage, but I have no idea what's under the hood.
3. The Fermi shows negative year-one take-home. What's the assumption on customer count that gets you to break-even, and what's the acquisition channel you'd actually use?

## Verdict: on-the-fence

The honesty is real and it's rare enough to hold my attention. But I still do not know if this is a product or a business plan for a product, and that confusion is the page's fault, not mine.

---
*Memo by skeptic persona, generated 2026-06-18. Studio breaks own self-grading loop.*