The editor side of Weft, a
programming language and framework for AI orchestration: as flexible as
an agent, as reliable as code. Open a .weft file and the extension
shows it as a live graph; run it, and the graph lights up with the
run as it happens.
Weft is made by Weavemind and developed in the open at github.com/WeaveMindAI/weft.
- Graph view. Every
.weftfile opens as an editable node graph next to the text. Edits flow both ways: change the text and the graph follows, change the graph and the text follows. - Projects panel. The projects registered with your local weft runtime, with run / stop / open right there.
- Runs panel. Past and running runs, each one inspectable node by node in the graph.
- Live diagnostics. The weft compiler checks your file as you type and reports through the Problems panel.
- AI streaming edits. SEARCH/REPLACE blocks streamed by an AI
chat apply to the open
.weftfile as they stream, with the graph updating.
The extension talks to the weft runtime on your machine. If you have
not installed it yet, read
the install guide;
it is one script. The extension asks the weft command for your install's
address (http://127.0.0.1:14111 for the default install) and its key, and
picks up the new address if the install restarts on another port. If you want
it on another install of this machine, set weft.dispatcherUrl to that
install's address, and the extension sends that install's own key.
It talks to the weft runtime on localhost and to nothing else: no
telemetry, no account, no outside service. To compile and run your
programs it starts the weft command that the install script put on
your PATH.
The full documentation lives at weavemindai.github.io/weft, and questions and bug reports are welcome on the issue tracker.
See LICENSE.