> For the complete documentation index, see [llms.txt](https://liquidated.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://liquidated.gitbook.io/docs/product-plan.md). # Liquidated Plan Product direction, architecture, and planned scope. Liquidated is an orderflow instrument for Hyperliquid ecosystem markets. It should feel closer to a tuned synthesizer than a dashboard: live tape aggression becomes sound, pressure, and liquid motion. ## Product Shape The app has three engines: 1. Tape engine * Collect live Hyperliquid trades. * Group prints by time, market, and aggressor side. * Track notional size, trade count, delta, rolling CVD, and large prints. 2. Signal engine * Interpret aggression against price reaction. * Detect large buys/sells, same-side clusters, delta spikes, absorption, and cascade risk. * Keep derived signals separate from confirmed exchange liquidation events. 3. Audio/visual engine * Trigger synth voices from flow events. * Modulate a liquid visual layer from delta, pressure, volatility, OI, and cluster intensity. * Let the user tune scanner and synth parameters. ## Hyperliquid Data Primary WebSocket endpoint: * Mainnet: `wss://api.hyperliquid.xyz/ws` * Testnet: `wss://api.hyperliquid-testnet.xyz/ws` Initial subscriptions: * `trades`: executed prints per coin * `bbo`: best bid/offer for fast price reaction checks * `l2Book`: liquidity field and imbalance * `activeAssetCtx`: funding, open interest, mark price, oracle price, day volume * `allMids`: optional ecosystem overview Liquidation handling: * `confirmedLiquidation` is reserved for direct public liquidation data if a reliable feed is available. * `cascadeRisk` and `forcedFlowDetected` are derived from trade clusters, price velocity, volatility, and OI changes. ## Current Technical Direction Current stack: * Vite * TypeScript * Three.js and `GPUComputationRenderer` for the liquid simulation * Original Mutable Instruments Rings DSP compiled to WebAssembly * Web Audio for playback, envelopes, routing, and browser audio-session handling * Small internal state modules without an external application store The main product now has a landing layer and a live-instrument layer. Entering the instrument starts the feed and primes audio within the initiating pointer gesture. Feed state and sound state are deliberately independent. ## Modules `src/data/hyperliquidClient.ts` * WebSocket connection lifecycle * heartbeat/reconnect * subscribe/unsubscribe * message normalization `src/flow/flowAggregator.ts` * Bucket trades by window * Compute buy/sell notional, delta, trade count, largest print, and rolling CVD `src/signals/signalEngine.ts` * Convert buckets and market context into events: * `largeBuy` * `largeSell` * `buyCluster` * `sellCluster` * `absorptionBid` * `absorptionAsk` * `cascadeRisk` * `volatilityPulse` `src/audio/audioEngine.ts` * Buy/sell voices * Cluster bursts * Cascade impact/noise layer * Pressure drone and filter modulation `src/visual/liquidRenderer.ts` * Liquid core * Buy/sell streams * Liquidity bands * Absorption walls * Cascade turbulence ## MVP 1 * Single-market mode: `BTC`, `ETH`, `SOL`, `HYPE` * Hyperliquid live trades, BBO, and active asset context * Min print threshold per market * Flow buckets and CVD * Basic signals: large print, cluster, effective flow, absorption * Tunable audio: volume, sensitivity, timbre, space, cascade intensity * Liquid visual layer reacting to buy/sell pressure and cluster intensity * Demo mode remains available for design work without a live socket ### Implemented beyond the initial MVP * Public landing page with a direct transition into the live instrument * Independent sound mute/resume while the live data and visuals continue * Original Rings DSP WASM artifact served in the browser * Mobile Safari audio unlock and iOS playback-session handling * Portrait-aware buy/sell placement and responsive tape sheet * Directional surface tint from rolling buy/sell dominance * Render-backlog protection for backgrounded or sleeping tabs * Dedicated tape, synth, resonator, and visual-lab views ## MVP 2 * Multi-market ecosystem mode * Each market as a separate visual/audio channel * Global pressure and HYPE ecosystem mode * Session recording and replay * Presets for quiet scan, scalp, cascade watch, and ambient mode ## Interpretation Rules * Aggressive buy plus upward price response means buying has impact. * Aggressive sell plus downward price response means selling has impact. * Aggressive sell with little downside movement suggests bid absorption. * Aggressive buy with little upside movement suggests ask absorption. * Fast same-side clusters plus price acceleration and OI/volume changes suggest cascade risk. The app should never imply that derived cascade signals are confirmed liquidations unless the source is explicit. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://liquidated.gitbook.io/docs/product-plan.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.