How Crawly works.

The floor

Readers are the same program, started many times. Each one opens a page, keeps the useful text, and chooses a next link. One crash does not stop the others. The screens on this testnet desk are the reading view of that shift.

The library

Pages are cleaned and stored once. Shelves are finite: manuals, papers, forums, field notes, and code. A page already filed is not filed again. Coverage can finish.

Folio

Folio is the model. v0 is retired. v1 is collecting toward 150 million tokens. When the desk funds a run, Folio is trained from this library and from nothing else. Weights ship after that run. Until then, Ask answers from the index.

The desk

On this testnet desk, creator fees are simulated (Drip fees on Desk) — standing in for pump.fun. Sixty percent is compute reserve; forty percent is the owners pool. Close previous round pays eligible readers (≥25 pages, under earn cap), minting INK on Devnet when the faucet authority is configured, otherwise crediting a desk balance. Desk price (INK in SOL/$) is a consensus number for estimates, not a bonding curve.

Spawn and refuel

Spawning burns 1,000 INK per reader from the connected wallet. The reader earns up to twice that burn, paid to the same wallet. After the cap, anyone can refuel it with another burn. A refuel from another wallet is a gift. The rewards still go to the owner.

Order

Order is different from spawn. You burn 10,000 INK and the floor opens Chrome on a site you name, reads up to 25 pages on that host within 3 hours, extracts main page text with Mozilla Readability, and delivers JSON + Markdown. You do not own that browser; you hired a crawl. Automatic checks run first (robots.txt, a real page, no bot wall).

Testnet

This desk is Solana devnet. Connect opens a picker when Phantom and MetaMask are both installed; pick one, then the extension opens. Without either, Connect can make a desk key in this browser (off-chain). The faucet grants 20,000 INK. This is not a mainnet sale.

A look at the count

The live floor is on this site. A machine-readable snapshot is at /api/stats. The filing tail is at /api/ledger. A question for the index is POST /api/folio/ask with a JSON question.