We're excited to announce Private Memories—a new memory tier that's yours. Where Team Memories are admin-authored facts shared with everyone, a Private Memory is private and author-only: only you see it, and it loads into only your own Claude Code sessions.
The Problem: You Keep Re-Explaining Yourself
You've told your AI a dozen times. "Keep answers terse." "I'm a backend dev—skip the frontend hand-holding." "I prefer the explicit version over the clever one." And every new session, it's forgotten. You start over.
Team Memories solved this for facts the whole team shares—the package manager, the error-handling convention, the deploy windows. But a lot of what makes your AI feel like yours isn't team-wide at all. It's personal. And you don't want to push your working style onto teammates or clutter the shared team context with it.
The Fix: A Memory Tier That's Only Yours
A Private Memory is a short, durable fact about how you like to work. It's injected into your Claude Code sessions at startup—and only yours. Teammates never see it. Even the workspace owner or admin can't see another member's Private Memories. They're private by construction.
Think of them as the small set of preferences you wish your AI just remembered about you, without ever bleeding into anyone else's session.
Two Ways to Create One
- Add one yourself. Open the Private Memories page in the app and write a fact—just like Team Memories, but scoped to you alone.
- Let the AI capture it. As you work, SuperClawd notices a durable personal preference and saves it for you. No prompt, no decision—it just shows up in your Private Memories, ready to edit or remove.
The Bigger Story: Memory Is Now Automatic
Private Memories arrive alongside a shift in how SuperClawd handles memory across the board: you never have to decide where something goes.
As you work, SuperClawd captures durable facts and routes each one to the right place on its own:
- A build command, a repo gotcha, "the integration tests need the Docker stack up first"—that's shared knowledge about the work, so it becomes a Repo memory the whole team benefits from.
- A personal working preference—"keep answers terse," "I'm a backend dev"—is about you, so it becomes a Private memory only you ever see.
You're never asked "where should I save this?" The capture happens quietly in the background, and the routing is inferred from what the fact is about: the work and the team go one way, the person goes another. A team convention phrased "I always…" still stays with the team—it never leaks into your private store.
Private by Construction
This is the part we care about most. Memory that watches you work could be creepy. We built it so it isn't.
- Zero LLM inference on our side. The judging that decides what's worth remembering runs on your own Claude—not on our servers. We don't run a model over your work.
- Content-free facts only. What's stored is a short, distilled preference—never your code, your diffs, your command output, or your conversation.
- Fully inspectable. Everything that gets captured is visible, editable, and removable in the app. Nothing is hidden.
- You're in control. Private Memories are yours alone—account-scoped, so they follow you into every workspace you join. You toggle them on or off with the private-memory setting in
superclawd config(on by default); no workspace owner can turn another member's on or off.
Why It Matters
Your AI stops re-learning who you are every session. The way you like to work follows you from one session to the next—without leaking to teammates and without polluting the shared team context that everyone relies on.
Team Memories keep the whole team aligned. Private Memories keep your AI feeling like yours. Together, the right context shows up in the right session, automatically.
Get Started
Private Memories are on by default and live on your account—not inside any single workspace—so they're ready wherever you work.
- Go to app.superclawd.com
- Open the top-level Memories page on your account
- Add a preference—or just keep working and let the AI capture one for you
- Your next Claude Code session picks it up automatically—and only yours
Want the full details on how memories are captured, routed, and injected? Read the docs at docs.superclawd.com.
Pro tip: Start with the one thing you find yourself re-explaining to your AI most often about how you work. That's your first Private Memory.