poi: A Plugin-Driven Browser for KanColle Players Who Want More Than a Tab

What poi Actually Solves

poi solves a specific problem: KanColle is a browser game that runs well in a normal browser, but serious players want more. They want to see fleet data without alt-tabbing, they want notifications when expeditions complete, and they want to automate repetitive actions. poi wraps the game in an Electron shell and layers on a proxy, a cache, and a plugin system. The target user is not a casual player. It is someone who tracks resources, sorts fleets, and wants the game client to do part of the thinking. The README calls it a scalable browser and tool set, and the emphasis on plugins and data synthesis makes that clear. If you only play KanColle occasionally in a regular tab, poi is overkill. If you play daily and manage multiple fleets, it becomes a control panel.

How the Proxy and Cache Work Together

The core mechanism is a local proxy. The README lists HTTP, Socks5, and PAC (Experimental) as supported proxy types. That proxy sits between the game client and the KanColle servers. Every request the game makes passes through poi, which allows the tool to cache responses and synthesize data. The custom cache feature means you can control what gets stored, which is useful for reducing load or for replaying certain responses during development. The data synthesis and analysis layer then takes the raw API responses and turns them into something a plugin can consume, like fleet composition or resource trends. This architecture is why poi can do things a normal browser cannot: it sees the traffic, not just the rendered page. The trade-off is that any change in KanColle's API format can break the synthesis layer, and the README does not detail how often that happens. It is a design that depends on the game staying stable.

The Plugin System Is the Real Selling Point

The README lists plugin support as a key feature and points to a wiki page for a list of available plugins. That is the extent of the documentation in the README. The plugin system is what makes poi more than a themed browser. Plugins can add UI panels, automate quests, or send notifications. The development docs are available in English at dev.poooi.app, which suggests the plugin API is a first-class citizen. The Electron and React stack means plugins likely interact with the UI through React components and Redux state, but the README does not specify the exact API. This is a gap. A player who wants a specific plugin must check the community list and hope it is maintained. The plugin ecosystem is community-driven, so quality varies. For an engineer evaluating this, the plugin system is both the strength and the risk: it gives flexibility, but you cannot rely on any single plugin being updated.

Getting It Running: Releases, Nightlies, and From Source

The easiest path is a GitHub release, which provides binaries for the major platforms. The README also lists community-maintained releases: AUR for Arch Linux, Homebrew Cask for macOS with brew update && brew install --cask poi, Chocolatey for Windows with choco install poi, and a Scoop bucket. These are unofficial, so they may lag behind the latest release. Nightly builds exist for Windows on Appveyor and for Linux and macOS at nightly.poi.moe, but the README warns that artifacts could be periodically emptied. That means a nightly link might 404 tomorrow. For developers, the from-source route is straightforward: git clone, npm install, npm start. This assumes Node.js and npm are present. The v12.0.0 release is from 2026-08-26, so the project is active. The choice between a stable release and a nightly is a trade-off between stability and freshness, and the README does not state which is more reliable.

Where poi Is the Wrong Tool

poi is not a general-purpose browser. It is built for one game. If you stop playing KanColle, the tool becomes irrelevant. The proxy and cache are tuned for that game's traffic, not for general web browsing. Another limitation is the experimental status of PAC proxy support. The README marks it as Experimental, which means you should not rely on it in a production setup. If you need PAC for a corporate network or a complex proxy chain, you may hit bugs. Also, the dependency on Electron means the application is memory-hungry compared to a lightweight browser extension. A player with an older machine might see performance issues. The README does not mention system requirements, so this is an inference from the Electron stack. Finally, the plugin list is external, so you cannot assess the quality of plugins from the README alone. You must visit the wiki and evaluate each plugin's maintenance status yourself.

Alternatives: Browser Extensions and Dedicated Viewers

The main alternative is to use a browser extension like KC3改 (KanColle Command Center) or similar tools that run inside a regular browser. These extensions also intercept game traffic and provide overlays, quest tracking, and expedition timers. The difference in approach is that extensions run within the browser's security model, so they cannot offer a custom proxy or a full local cache. They also cannot provide a standalone window or system-level notifications as easily. Another alternative is to run the game in a dedicated browser profile with userscripts, which is lighter but offers less integration. poi's advantage is that it is a self-contained application with its own proxy and plugin runtime. The disadvantage is that it is a heavier install and requires trusting an Electron app with your game credentials. For a player who wants minimal footprint, an extension is simpler. For a player who wants deep automation and data analysis, poi's architecture is more capable.

Maintenance, License, and the Icon Caveat

The project is under the MIT License, which is permissive for code reuse. However, the README includes a specific note: the software icon and SVG icons are not licensed by MIT and cannot be used in projects not associated with poi. That is a real constraint if you plan to fork or redistribute. The project has recent activity, with v12.0.0 released on 2026-08-26, so it is not abandoned. The maintenance model relies on community contributors and backers, as shown by the contributors and backers sections. The upgrade cost is tied to the plugin ecosystem: when poi updates, plugins may need updates too. The README does not provide a migration guide, so you must test plugins after each major release. The nightly artifacts being periodically emptied means you cannot rely on a specific nightly for long-term use. For an organization, this is not a tool you would deploy widely; it is a personal tool with a community support model.

Editorial conclusion

Adopt poi if you are an active KanColle player who wants a dedicated client with plugin-driven automation, proxy flexibility, and local data analysis, and you are comfortable managing a tool that relies on community packages and nightly builds. Do not adopt it if you expect a polished, officially supported installer on every platform or if you need a general-purpose browser. Before adopting, verify that the plugin you need is compatible with the current v12.0.0 release, check the community release source for your OS (AUR, Homebrew, Chocolatey, or Scoop), and confirm that the nightly artifact you use is fresh, since the README warns artifacts can be periodically emptied.

Official sources

1. Official documentation
2. Official README
3. Project repository
4. Release notes