Memex

Most AI assistants don't know anything about you. Every new conversation starts from zero — they don't know what you've been reading, what links you've saved, what notes you've written, or what files you've downloaded.

Memex fixes that. It's a background daemon that runs on your Mac and passively indexes your digital activity into a local database — made available to any LLM through standard interfaces. No cloud storage. No third-party servers. Everything stays on your machine.

TypeScriptSwiftMCPOpenAPI 3.1.0Cloudflare Tunnel

What it captures

Screenshots— OCR'd locally using Apple's Vision framework (Swift)
Clipboard — monitored continuously; YouTube URLs auto-fetch transcripts, Twitter links scrape post content
Downloads — PDFs, markdown files, CSVs parsed and indexed on arrival
Apple Notes — synced periodically via AppleScript
Links from your phone — shared directly from Android via HTTP Shortcuts → POST to local API

How it works

[Screenshots / Clipboard / Notes / Downloads]
              ↓
        Memex Daemon (Node.js)
              ↓
     Local JSON index (db.json)
       ↙              ↘
 MCP Server          HTTP Server :4322
 (stdio)             (Bearer auth)
     ↓                    ↓
Claude Desktop      Cloudflare Tunnel
                         ↓
                   ChatGPT Mobile /
                   Android Share Sheet

The database is a flat JSON file written atomically to disk. Search runs through a custom TF-IDF model — no embeddings, no API calls, fully offline. It weights keyword relevance, recency, document type, and URL path matches to return the most relevant result for any query.

Two interfaces serve the database simultaneously: a local MCP server (stdio transport) for Claude Desktop and MCP-compatible editors like Cursor, and an HTTP server exposed over a free Cloudflare tunnel for ChatGPT Custom Actions and mobile ingestion.

Things that were annoying to figure out

ChatGPT's 1MB tool result limit

Returning full PDF text crashed the action with a ResponseTooLargeError. Fixed by truncating all returned content to 2,000 characters server-side.

Claude Desktop's MCP connection dropping on startup

Happened because the HTTP server was initializing before the MCP transport was ready. Fixed by sequencing the boot order and adding retry logging.

Dynamic pages that block scraping

cosmos.so, Instagram, login-walled portals — @mozilla/readability extracts nothing from them. Rather than silently failing, the system now writes a URL stub with the timestamp and source URL so the LLM can still confirm the link was saved.

Android doesn't have Universal Clipboard

iCloud clipboard sync only works on Apple devices. Solved by routing phone shares through HTTP Shortcuts → Cloudflare tunnel → local ingest API instead.

Why certain things were built the way they are

Flat JSON over SQLite

The dataset is small and mostly read-heavy. Atomic file swaps give crash safety without adding a DB engine. The file is also human-readable and trivially portable.

TF-IDF over embeddings

Embeddings need either a local GPU or an API round-trip. Both break the offline-first constraint. TF-IDF runs in microseconds and works well for personal context retrieval where queries tend to be specific.

Cloudflare Tunnel over ngrok

No signup, no account, no bandwidth limits, and the URL persists for the session. Free ngrok rotates URLs on restart and throttles connections.