Cross-surface memory. Save a fact, a decision, a snippet of context in Claude Code. Recall it later in Claude.ai chat — or vice versa. One substrate, every Claude surface.
Exact, not approximate. Most "AI memory" tools run the LLM over your data on both write and read, so recalls drift. MinSets stores the exact text and returns the exact text. The model only interprets — it never rewrites what you saved.
Yours, isolated. Sign in with GitHub, get a personal token, and your memory is yours alone — no other user can see it. MinSets never uses your content to train AI, run embeddings over it, or share it with any third party. Full data posture: /privacy.
1. Sign in with GitHub → copy your token.
2. Drop it into your Claude Code MCP config (we give you the exact snippet).
3. Save something. Recall it from another Claude surface. Done.
Full steps in the install guide.