MindFile - your AI memory od Jwum

Every AI tool starts from zero. You re-explain your job, your stack and how you want answers written - then the next tool has never heard of you.

MindFile gives you one memory that follows you. Save it once and MindFile adds the relevant parts to the first message of each new chat on ChatGPT, Claude, Gemini, Grok, Perplexity, DeepSeek, Le Chat and Copilot.

STORED ON YOUR DEVICE. NOT OURS.
There is no MindFile account, no sync server and no telemetry. Your memories live in your browser's own storage, on your computer. Other tools in this category keep your profile in their cloud - that is precisely the part we removed.

IT SENDS WHAT FITS, NOT EVERYTHING
MindFile ranks your memories against what you are actually asking about and shares only the ones that fit. Matching runs on your machine: an on-device model scores relevance locally, so nothing is uploaded to work out what is relevant. Prefer the old behaviour? One setting shares everything, every time.

IT NOTICES WHEN YOU CONTRADICT YOURSELF
Memories drift. You switch tools, move city, change your mind. MindFile spots two memories that disagree and holds BOTH back rather than feeding a chat something stale - then shows you the pair side by side so you can say which one is current. Nothing is deleted without you.

A Memory Power score tells you whether your memory is in good shape: fresh, free of duplicates, and free of unsettled disagreements.

YOU ARE ALWAYS IN CONTROL
A chip shows how many memories are about to be shared, before you send. One click skips them for that chat. Auto-inject can be switched off per site. The context block is plain, labelled text at the top of your own message - never hidden, never rewritten.

FREE
- 30 memories
- Smart recall on all 8 supported sites, including on-device relevance matching
- Conflict detection, so contradictions never reach a chat unnoticed
- Save any text in one click - select it in a reply, or add it from the popup
- Import a memory pack somebody shared with you
- Guided setup to fill in the basics

PRO
- Unlimited memories
- Multiple profiles, so work and personal context never mix
- Ask your memory: ask a question and get an answer drawn from your own notes, with each memory it used shown as a source you can check
- Have your AI settle contradictions for you, and suggest which memory is current
- Extraction: MindFile reads a conversation and proposes memories, and you approve each one before it is saved
- Create memory packs to share a set of memories as a single file
- Export your whole vault as one portable file

BRING YOUR OWN KEY, AND WE NEVER METER IT
Every AI feature runs on an API key you provide - OpenAI, Anthropic, Gemini, or a local Ollama model. There is no per-message credit, no monthly allowance and no cap on top of the key you are already paying for. Your key talks directly to the provider you chose; it is never sent to us.

That is also why the free tier can stay genuinely useful: we are not paying for anyone's inference, so we do not have to ration it.

KEEPS WORKING WHEN THE SITES CHANGE
When a chat site redesigns its page, extensions that hardcode its layout break until they ship an update and wait out a review. MindFile loads its per-site definitions as a signed data file, verified against a key built into the extension, so a broken site can be fixed the same day without a new release.

On Firefox, site access is opt-in. After installing, Firefox will ask you to allow MindFile on each chat site you want it to work on, and it stays completely dormant until you do.