memorised-them-all

Memorised them All

Give your AI assistant a private, living memory of your files — without paying for it in tokens.

Point it at a folder of PDFs, Word/Excel files, or whole archives. It reads and remembers them entirely on your own computer — no AI models to download, no copy-pasting, no uploads, no API keys, no surprise token bills. Then just ask — in Claude, Gemini, Cursor, VS Code, Windsurf, ChatGPT's Codex, Grok, and more. Update it in seconds, and move or merge memories between machines.

_{v3.2 · works with Claude · Gemini · Cursor · VS Code · Windsurf · Codex · Grok · 100% local · deterministic · model-free · token-free · incremental · portable · English + Bengali · what's new →}

What is this? · Why it's different · Get started · First memory · Living memory · Bengali · FAQ · Advanced

🧠 What is this?

Imagine handing your AI a filing cabinet of your documents and saying "remember all of this." Later you just ask questions, and it answers from what it remembers — telling you which document each fact came from.

That's Memorised them All. It's a small add-on (an MCP server) that works with Claude, Gemini, Cursor, VS Code, ChatGPT's Codex, Windsurf, Grok, and more (one command sets them all up — see below). It:

1. Reads your files — PDFs, Word/Excel/PowerPoint (including old .doc/.ppt/.xls), web pages, CSVs, EPUBs, even zip/rar archives (unpacked safely) — and converts them to clean text on your computer.
2. Builds a memory — a searchable map of the people, topics, and facts inside them (a "knowledge graph"), plus a tidy synopsis and per-document notes.
3. Lets you ask — your assistant recalls just the relevant, cited snippets when you ask, instead of you pasting whole files into the chat.
4. Stays current and portable — re-running only re-reads what changed, and you can export a memory, move it to another machine, or merge several into one.

The one-line idea: AI context tokens cost money; your computer's effort is free. So all the heavy lifting happens locally, and the assistant only ever receives a tiny answer. Memorising a 500-page folder costs roughly zero chat tokens.

💬 See it in action

Once it's installed, you just talk to your assistant normally:

You:   Memorise everything in ~/Documents/contracts.
AI:    ✅ Digested 38 files → 421 facts across 7 themes. (took ~30s, all local)

You:   Which contracts mention an auto-renewal clause, and when do they renew?
AI:    Three do — the Globex MSA (renews 1 Jan, 60-day notice), … [cites each source]

You:   I added two contracts and deleted one — refresh the memory.
AI:    ✅ Updated: 2 added, 1 removed, 36 unchanged (re-read only the 2 new files).

You:   What changed between the 2023 and 2024 progress reports?
AI:    Three headline shifts — … [cites each source document]

Nothing left your machine. The assistant never saw the 38 files — only the small, cited answers. Behind the scenes, recall uses a fast on-device keyword search (BM25) to pick the few most relevant snippets, and declines when your question isn't actually covered.

🌟 Why it's different