The Moxie Library

Description and use

Mona Andrei built a library that consolidates 710 posts over 16 years, enabling retrieval of related texts while drafting new articles.

Problem

Mona Andrei needed all her 16 years of published writing in one place and a way to be pointed back to what she had already written when starting new content.

Outcome

The library aggregates 710 pieces, surfaces related past passages during drafting, and visualizes all work as a graveyard. It improves internal linking and helps Mona write with awareness of her own corpus.

Workflow

• Collect public posts from WordPress blog (650), Westmount Magazine (37), and Substack (23).
• Filter Westmount posts to those authored by Mona Andrei.
• Cut each piece into ~120‑word passages (2,809 passages).
• Embed passages using all‑MiniLM‑L6‑v2 via transformers.js in WebAssembly.
• Store 2,809 vectors (384 numbers each) as 1.08 MB in browser memory.
• When drafting, cut draft into passages, embed, and score against stored vectors to find closest past passages.
• Retrieve the closest past piece, its quoted passage, publication info, and copy‑link button.
• Display results on the desk; show visual graveyard of all pieces with tombstones or books based on internal linking.
• Provide copy‑link button for WordPress/Substack.
• Set robots‑noindex meta to prevent search engines from indexing the library.
• Deploy on Vercel.
• Tested by typing sample sentences and verifying returned passages.

Evidence

She wanted two things: everything she has written in one place, and something that would point her back at what she already wrote when she sits down to write something new.

Source

I built her a social media scheduler. It had a landing page and a logo and it ran on sample data. She found a product online that already did that, and texted me: "Found it. I will find the hole."

Source

This weekend I built what she asked for the first time. It is called The Moxie Library, and it is a single web page.

Source

When she pauses typing, the draft is cut into passages the same way, each one is embedded in the tab, and every piece is scored by its closest passage.

Source

I built her a social auto‑poster that month. It was the wrong thing, and it is not in this repo. This is the thing she asked for.

Source

The model runs through transformers.js on WebAssembly, in a web worker, inside her browser tab.

Source

The page asks search engines not to index it, so it does not compete with her own blog.

Source