SkillAgentSearch skills...

Papertrench

Paper-trade Solana memecoins on the sites you already use. Real prices, fake money, and a record you can actually learn from.

Install / Use

npx skills add OnlyTerp/papertrench

Installs into whichever agent you are using.

About this skill

Quality Score

0/100

Supported Platforms

Universal

README

<div align="center"> <img src="docs/assets/banner.svg" alt="PaperTrench" width="100%">

Paper-trade Solana memecoins on the sites you already use. Real prices. Fake money. A record you can actually learn from.

License: MIT Tests Chrome MV3 No tracking

</div> <div align="center">

▶ Watch the launch video

<a href="docs/video/papertrench-launch.mp4"><img src="docs/assets/preview.gif" alt="PaperTrench launch video preview" width="82%"></a>

</div>

What it is

PaperTrench is a Chrome extension that overlays a paper-trading terminal on Axiom, Padre, Photon, GMGN, BullX, Dexscreener, Birdeye, Jupiter, Pump.fun, Fomo, and Lute. You trade the real chart, at the real live price, with money that isn't real — then review exactly what you did and why.

It exists because the usual way people learn this market is to lose money finding out that they chase, oversize, and round-trip their winners. This tells you that in an afternoon instead.

<div align="center"> <img src="docs/assets/overlay.png" alt="The PaperTrench overlay on a live chart" width="86%"> </div>

Why it's different

It never invents a price. Every fill uses a price validated against a trusted anchor. When no source has a price yet, the panel says so instead of showing a number.

It measures process, not just P&L. Anyone can get lucky. PaperTrench grades how much of each move your exit actually captured, how big your positions were relative to your book, and whether you followed the plan you wrote before entering.

It works on brand-new launches. Dexscreener doesn't index a coin until it has seen a pool — measurably after the coin exists. PaperTrench queries Jupiter in parallel and anchors off whichever resolves first, so a coin seconds old is tradeable.


Features

Trading

| | | |---|---| | Live overlay | Quick-buy presets, % sells, and a P&L that updates on the site's own price feed — event-driven, not polled. | | Positions bar | Every open position pinned to the top of the page with live P&L. Click one to jump to its chart. | | Native chart fills | On Padre, your paper buys and sells render as real TradingView markers with average fill/exit lines. | | Armed buys | Click buy before a brand-new coin has a quote and the order arms, then fills the instant the first trusted price lands. | | Instant X links | Opt-in: clicking an X post or profile on a trading site opens it in a kept-warm viewer tab in ~0.5s instead of a ~3.5s cold tab — and hovering a link first prefetches it, making the click near-instant. Costs one muted background x.com tab while enabled; Ctrl/Cmd/middle-click still opens normal tabs. | | X-Ray | Opt-in: land on any X profile or post and the intel card is already there — account age, bio / display-name / @handle changes, every contract address the account has posted, and Smart Following (its biggest followers). Built from the X page's own data, on your machine. It states what it cannot know: change history starts the first time your device sees an account, and the card prints that date. |

Learning

| | | |---|---| | Trade thesis | Write why you're taking a trade while the position is open — before you know the outcome. Tag the setup, set a target and stop. | | Exit quality | Every round graded excellent / sold early / round-tripped, with the SOL you left on the table quantified. | | Position sizing | Average and largest trade as a share of your book, flagging anything over 25%. | | Setup performance | Win rate and average P&L per setup tag — so revenge can be compared honestly against narrative. | | Session replay | Scrub through any trade: every fill, chart frames, and the screen recording if you made one, all on one timeline. | | Shareable P&L cards | Turn any closed round into a posting-ready image, with your own picture or GIF behind it. | | AI coaching | Optional review of a round or a whole session through any OpenAI-compatible endpoint. Judges process against your stated plan. |

Verifiable record

Every fill is committed to a hash chain the moment it happens, which makes three things checkable: ordering (no inserting or deleting trades afterwards), pre-commitment (no backdating a winner), and price reality (each fill records mint, price, and timestamp, so it can be re-checked against real history).

The dashboard re-derives your result from the chain and shows it beside the displayed number. If they disagree, the stored state was edited.

Stated plainly: this is evidence, not proof. Anyone can run modified code locally. A leaderboard built on this must recompute standings server-side from the submitted chain — never from the number the app displays. See docs/LEADERBOARD.md for the protocol.


Install

PaperTrench is not on the Chrome Web Store yet. Loading it unpacked takes about thirty seconds.

  1. Download — grab the latest release zip and unzip it, or clone this repo.
  2. Open chrome://extensions and turn on Developer mode (top right).
  3. Click Load unpacked and select the folder that contains manifest.json — the unzipped release folder itself, or the extension/ folder if you cloned the repo.
  4. Pin PaperTrench so the popup is one click away.

Getting "Manifest file is missing or unreadable"? The folder you selected doesn't have manifest.json at its top level. Open the folder you unzipped and select the level where manifest.json sits.

Open any supported token page and the overlay appears within a second or two.

<div align="center"> <img src="docs/assets/dashboard.png" alt="The PaperTrench dashboard" width="86%"> </div>

First trade

  1. Go to a token on any supported site, e.g. https://trade.padre.gg/trade/solana/<mint>.
  2. Wait for the panel's status dot to turn green — that means a trusted price is live.
  3. Pick a SOL amount and hit BUY (PAPER). Your balance drops; a position card appears.
  4. Write your thesis while it's open. Why this trade? What's the target?
  5. Sell 25 / 50 / 75 / 100% when you're ready.
  6. Open the dashboard to see the round graded — and what you left on the table.

P&L cards

Every closed round has a Share button that opens a card composer. Drop in any picture or GIF as the background — it's cover-fitted and scrimmed so the numbers stay readable over a busy image — then download a 1200×675 PNG ready to post.

The numbers on the card are read straight from the engine's own round record, so a shared card cannot show a result your journal doesn't contain.

The composer now works like the terminals people know: a background gallery (built-in looks plus your own uploads, saved for next time), a Customize panel for which stats show, and Download / Copy to clipboard for paste-and-go posting. Two things are always drawn and cannot be customized away: the PAPER watermark across the middle, and the full-width PaperTrench brand bar with "paper trading — not financial advice" and the project URL. The branding is drawn by a code path that reads no settings at all — there is no combination of options that removes it. A simulated trade should never be passable as a real one.

Privacy

  • No account. No signup. No telemetry. Nothing is sent anywhere about you.
  • A real off switch. The ⏻ button in the popup turns PaperTrench fully dormant — nothing injected or rendered on any site, live in every open tab — until you turn it back on. Wallet and settings are kept.
  • Everything is local. Trades, settings, replays, and recordings live in your browser's own storage.
  • Network calls are only: the public Dexscreener and Jupiter price APIs, X's public oEmbed endpoint if you opt into hover tweet previews (no login, dnt=1, posts you hover only), and — only if you configure it — your own AI endpoint.
  • Recordings never leave your machine. They're stored in IndexedDB and saved to your downloads folder.

The permissions in manifest.json are the minimum needed: storage for your wallet, tabs/activeTab for chart frames, and offscreen for the optional recorder. Content scripts run only on the supported trading sites — never anywhere else. The full permission-by-permission audit lives in docs/PERMISSIONS.md.


Tests

cd extension
node --test

872 tests covering price resolution, tick validation, portfolio arithmetic, the Padre chart bridge, fresh-launch handling, the positions bar, session replay, the attestation chain, and browser-context loading.

The suite is mutation-tested: fixes were verified by reverting them and confirming the tests fail. The one test that hits a live API skips — rather than fails — when offline.


Architecture

extension/
  manifest.json      MV3 manifest
  sites.js           per-site adapters (URL → token identity)
  quote.js           pure logic: pair selection, tick validation, P&L math
  resolver.js        Dexscreener + Jupiter lookups (raced)
  engine.js          portfolio engine: positions, fees, rounds, thesis, grading
  attest.js          tamper-evident fill chain
  replay.js          session replay model
  recordings.js      IndexedDB store for screen recordings
  price-bridge.js    MAIN-world hook into the site's own price feed
  content.js         Shadow-DOM overlay
  background.js      service worker: AI proxy, recording, frames
  dashboard.html/js  journal, rounds, replay, leaderboard, coach, settings

The decision logic is deliberately kept in pure functions (quote.js,

Related Skills

View on GitHub
GitHub Stars30
CategoryDevelopment
Updated5h ago
Forks8

Languages

JavaScript

Security Score

95/100

Audited on Aug 8, 2026

No findings