This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
An MCP server that exposes a Rubik's cube to an agent, plus a small dashboard to watch agent runs in the browser.
uv sync
make build-web
make serve # dashboard at http://localhost:8000
make mcp # MCP server (stdio)Related MCP server: Chess MCP Server
Five uv workspace packages under packages/: cube, observe, mcp_server,
harness (stub), dashboard. See docs/architecture/LAYOUT.md for the
single source of truth and docs/design/mvp.md for the framing.
Run logs live in $XDG_DATA_HOME/rubiks-cube-mcp/
(default ~/.local/share/rubiks-cube-mcp/), never in the repo.
The solve_hint tool wraps kociemba,
which ships an x86_64 C extension. On machines where that wheel doesn't load
(notably arm64 macOS without a matching wheel), kociemba transparently falls
back to a pure-Python implementation of the same algorithm. We use that
fallback path by default — it works on every architecture with no build step
and solves in ~1 second.
Want the native C extension speed? Run make solver-fast once. That
rebuilds kociemba from source against your local architecture (sub-millisecond
solves afterward). Requires a C compiler on your PATH — on macOS install
with xcode-select --install; on Linux apt install build-essential; on
Windows install MSVC build tools. Purely opt-in; nothing else needs to change
in our code, our wrapper picks up the C extension automatically when it loads.
This server cannot be deployed
Human-in-the-loop control panel for AI agents: start runs, review them, answer their requests.
Build, validate, and deploy multi-agent AI solutions from any AI environment.
Control a real browser from AI agents: navigate, click, type, screenshot, read the page.
Web infrastructure for AI agents: scrape, crawl and search the web, run browsers and agents.
