Website Summary: Neon Jianghu
This website is a highly stylized, cyber-noir barter scene set in a futuristic Chongqing-inspired world. It presents a fictional black-market negotiation interface where the player interacts with an NPC merchant named Brother Wei. The visual design is the dominant feature: heavy neon effects, animated rain, scanlines, glitch typography, floating sparks, HUD elements, and a game-like barter module create the feel of an interactive narrative demo or UI concept for a cyberpunk role-playing game.
Loaded SRC_URL
https://producingtechnology.com/65-apps/xionghaochen_179277_15200508_neon_jianghu.html
Summary of App Behavior
The app loads directly from embedded data rather than fetching a remote JSON file at runtime. It displays a merchant scene labeled HONGYADONG L4 with environmental flavor such as heavy rain and neon flicker. The top HUD shows the player wallet, including digital credits and scrap. The main content is split into panels for the merchant, inventory, and bargaining controls.
The merchant panel introduces Brother Wei as a live LLM-style vendor with an emotional state composed of temperament, mood, and stress level. Dialogue is shown with a typewriter effect. The inventory panel lists items with condition, base value, and marked-up asking price. Selecting an item updates the haggle module.
The haggle panel allows the player to move a price slider, see a computed success probability, and choose among three actions: haggle, pay asking price, or walk away. The app updates credits, stress, mood, dialogue, visual indicators, and toast notifications based on the result. Additional visual effects include click sparks, shake feedback on bad interactions, an animated rain canvas, and a scrolling footer ticker.
Things That Did Not Work as Expected
- The app presents itself as a live LLM merchant experience, but the merchant behavior is not actually driven by a live language model. Dialogue is selected from a small predefined set of scripted lines, so the experience is closer to a simulation than a true conversational bargaining system.
- There is no actual backend, persistence layer, or inventory mutation. Buying an item reduces credits, but the purchased item is not removed from stock, reserved, or tracked in a player inventory.
- The bargaining system gives the impression of deep intelligence, but acceptance is determined by a probability calculation and random roll rather than richer negotiation logic, memory, or context-aware dialogue.
- The loaded source URL and the uploaded file do not show an external data fetch. All scenario data is embedded directly in the page, so the world state is static unless the source code is edited.
- The emotional model is visually effective but mechanically shallow. Stress and mood change in limited ways and do not meaningfully alter inventory, faction behavior, or long-term consequences.
- Accessibility may be weaker than expected because the design relies heavily on glow effects, animation, color-coded cues, and pointer interaction. There is little evidence of keyboard-first support, semantic control labeling, or reduced-motion handling.
- The app is strong as an atmospheric prototype, but it does not yet feel like a complete trading game loop. There is no save state, no progression system, no branching narrative, and no multi-scene navigation.
Best Attempt at a Prompt to Improve the App
Improve this app into a more complete cyberpunk bargaining experience while preserving the current neon Chongqing visual identity. Keep the existing aesthetic, animations, and panel structure, but replace the scripted vendor behavior with a more dynamic negotiation engine that can respond differently based on item category, prior offers, player reputation, faction alignment, and merchant mood history. Add persistent game state so purchases remove items from stock, credits remain updated across sessions, and acquired items are stored in a player inventory. Expand the merchant system so stress and mood influence not only acceptance probability but also dialogue tone, price ceilings, refusal thresholds, item availability, and security escalation. Add additional vendors, randomized stock, item rarity, faction-specific rules, and consequences for repeated lowballing. Improve accessibility with keyboard navigation, ARIA labels, stronger semantic structure, and a reduced-motion mode. Add optional sound design, ambient audio, and subtle haptic-style feedback cues. Finally, introduce narrative hooks, reputation tracking, and scene transitions so this feels like part of a larger interactive world rather than a single static barter screen.