innercontext.com

What It Is

Inner Context’s core product, WaterCooler, is a local 3D visualization interface for AI Agent workflows. It does not create messages or run model inference itself. Instead, it reads SQLite databases written by OpenCode-related plugins and renders Agents, messages, unread states, and activity status as an “office”: each Agent sits at its own desk, messages are connected by arcs of light, red indicates unread, and green indicates read.

Core Capabilities

Its AI capabilities come from the accompanying OpenCode plugin ecosystem: opencode-coworker is used to create persistent AI coworker sessions, opencode-mailbox provides asynchronous messaging across sessions, opencode-cron supports sending scheduled tasks to sessions, and opencode-status tracks states such as thinking, tool execution, and idle. WaterCooler visualizes these states and lets users view an Agent’s full message history, send messages to a single Agent or broadcast to all Agents, and handle decision buttons with a choices field.

Pricing and Deployment

The materials indicate that all Inner Context software is released as open source under the MIT License, including WaterCooler and the four OpenCode plugins. No paid plans were found. Installation is via npm install -g watercooler or npx watercooler, and Node.js 18+ is required. The CLI can specify paths for the mailbox, coworkers, and status databases, as well as the port and host.

Pros and Cons

Its strengths are that it runs locally, keeps data on the user’s machine, and has clear privacy boundaries. The components have well-defined responsibilities, making it suitable for observing complex multi-Agent communication. The MIT license also makes secondary development easier. The limitations are also clear: it depends heavily on OpenCode and the SQLite database structure, and is not a general-purpose Agent platform by itself. It binds to 0.0.0.0 by default, so LAN access needs to be restricted manually. We did not find evidence of a Chinese interface, commercial support, permission management, or capabilities for large-scale scenarios.

Who It’s For and Access from China

It is better suited to developers, AI Agent tooling researchers, and users building local multi-Agent collaboration with OpenCode. It is less suitable for enterprise users looking for an out-of-the-box SaaS, team permissions, and managed operations. Access from China is not specified in the materials; because installation depends on npm, real-world usability may be affected by network access to npm, GitHub, or jsDelivr. Comparable alternatives include LangGraph Studio, CrewAI, AutoGen Studio, Dify, and Flowise.