WeaveMind

This platform is built around Weft, a programming language designed for AI to write and for people to read as a graph. The idea is to orchestrate AI systems as ordinary programs instead of building free form agents, so that the result is as flexible as an agent but as reliable as code. Each Weft program has two views that stay in sync, the code and a graph of its steps. Tangle is the coding assistant built into Weft, and it runs inside the tools a developer already uses, including Claude Code, Cline, Codex, Cursor, Devin Desktop, Gemini CLI, GitHub Copilot, Junie, Kilo Code and OpenCode. A user describes what they want in plain words and Tangle states the plan first, writes the prompts and any missing building blocks, tests every piece, builds a frontend if asked and deploys the program to the user's own cloud. When a case fails in production, Tangle can be given that case, work until it passes, and freeze the run so later runs show what changed. Weft also supports scoped runs, where any node or group can be started on its own from made up values, a saved real input or connected live services. Spending rules can route orders by amount, for example buying directly under a set limit and asking a teammate to approve larger amounts. Weft is free and open source and runs on the user's machine. The site states a price of $0 with no plans, seats or usage meter for the self hosted version. It describes itself as early, though it runs most projects, and cloud deploy works on GCP today.