Analysis updated 2026-05-18
Run a local shared room where you and an AI agent work on files together outside of chat.
Connect an MCP agent bridge so an automated agent can act inside the room.
Prototype a canvas style collaborative interface using the included React app.
Study the architecture docs to learn how the room hub, WebSocket, and SSE APIs fit together.
| iljazakharov/agentworkspace | abdulkader-safi/obsidian-image-magick-plugins | acartag7/mcp-sso | |
|---|---|---|---|
| Stars | 8 | 8 | 8 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 3/5 | 2/5 | 4/5 |
| Audience | developer | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Some integrations require filling in your own environment variables and API keys.
AgentWorkspace is a browser based shared room where a person and one or more AI agents can work together in the same space, instead of talking back and forth through a messy chat log. The room supports chat, canvas style objects you can arrange visually, shareable room links, and optional voice and agent integrations. The README describes this repository as an early demo snapshot, and it notes things might still be rough since the project is being actively updated as it stabilizes. What is actually included here is a Node.js server that runs the shared room and exposes it through regular web requests, WebSocket connections, and server sent events, a simple local JSON file used to store room state during development, a React and Vite based interface found in the app folder, an MCP server and agent bridge so automated agents can join and act inside a room, and a set of backend, frontend, and browser tests. Example environment files are included with blank placeholder values only. The README is explicit about what has been deliberately left out of this public copy: internal logs of how agents actually behaved, any real runtime databases or cached room data, real API keys or invite links, and private notes from the original development team's own workspace. This confirms the repository is a cleaned up snapshot of a larger private project rather than the full thing. To try it, you install dependencies and run a start command for the Node server, which listens locally on a specific port, and separately install and run the React app for its own development server. A handful of npm scripts are available for running tests, linting, a release check, and starting the MCP server. Some integrations need environment variables, which you set by copying the example file and filling in only the providers you actually plan to use, being careful never to commit real keys. The README does not state a license for this project.
A browser based shared workspace where a person and AI agents collaborate in the same room with chat, canvas objects, and voice options.
Mainly TypeScript. The stack also includes Node.js, TypeScript, React.
No license information is given in the README.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.