Zurück zu den Skills
windsurfinabox
von pfcoperez
Windsurf—Noch nicht bewertet
Deine Bewertung
Kurzbeschreibung
Windsurf's Cascade agent within a Docker image, to be used in headless mode
Beschreibung
# Windsurf in a box

Windsurf's Cascade agent within a Docker image, to be used in headless mode. With this image, you can use Windsurf Cascade agent without a Desktop environment which makes it possible to integrate in automatic processes such as code management pipelines.
## Usage
- Build the image:
```bash
docker build . -t windsurf
```
- Prepate your workspace directory with read and write permissions for user with `UID:GID=1000:1000`.
- Create a file named `windsurf-instructions.txt` and add it to the directory. It should contain your prompt for Windsurf.
- Get your Windsurf Auth token from a normal Windsurf editor:
- `ctrl+shift+p`
- Run the command Provide auth token:

- Run the image generated with this project:
```bash
docker run --rm -it --name windsurf -e WINDSURF_TOKEN=$WINDSURF_TOKEN -v /home/pablo/.config/Windsurf:/home/ubuntu/.config/Windsurf -v /home/pablo/temp/workspace:/home/ubuntu/workspace windsurf
```
You might have noticed that the token is passed as the `WINDSURF_TOKEN` and the workspace mounted with `-v /home/pablo/temp/workspace:/home/ubuntu/workspace`.
🚧 Right now, `-v /home/pablo/.config/Windsurf:/home/ubuntu/.config/Windsurf` is required for the local configuration setting the editor to accept suggested actions automatically. This is just temporary and next commit to this repo will remove the need of passing a pre-set configuration.
## How does this work?
This image leverages Xvfb to start a virtual X11 screen where the editor is launched, the Docker image entrypoint script uses `xdtool` to send commands as if it was the UI user. These commands include log-in with the provided token and the activation of a [preconfigured workflow](https://github.com/pfcoperez/windsurfinabox/blob/main/src/workflows/entry-workflow.md) with the following prompt:
> Follow the instructions described in the file windsurf-instructions.txt; Append a description of each step you follow in realtime and its outcome to the file windsurf-output.txt creating it if missing and including timestamps for each entry; Once finished, append a last line in windsurf-output.txt with the text "WORK-COMPLETED", also including a timestamp as with other entries (same format).
## Example
```
echo "write a openscad program to generate a merger sponge" > windsurf-instructions.txt
export WINDSURF_TOKEN="<REDACTED>"
docker run --rm -it --name windsurf -e WINDSURF_TOKEN=$WINDSURF_TOKEN -v /home/pablo/.config/Windsurf:/home/ubuntu/.config/Windsurf -v /home/pablo/temp/workspace:/home/ubuntu/workspace windsurf
```
After a while (<5 mins)...
```bash
cat ~/temp/workspace/menger_sponge.scad
```
```openscad
// OpenSCAD program to generate a Menger sponge
// The Menger sponge is a fractal cube with holes
// Function to create a single iteration of the Menger sponge
module menger_sponge(size, level) {
if (level == 0) {
// Base case: just a cube
cube(size, center = true);
} else {
// Recursive case: create 20 smaller cubes (27 - 7 removed cubes)
new_size = size / 3;
for (x = [-1:1]) {
for (y = [-1:1]) {
for (z = [-1:1]) {
// Skip the center cube and the 6 cubes that share a face with the center
if (abs(x) + abs(y) + abs(z) > 1) {
translate([x * new_size, y * new_size, z * new_size])
menger_sponge(new_size, level - 1);
}
}
}
}
}
}
// Parameters
size = 100; // Size of the initial cube
iterations = 3; // Number of recursive iterations
// Generate the Menger sponge
menger_sponge(size, iterations);
```
```bash
cat windsurf-output.txt
```
```
[2025-05-27T15:28:30Z] Starting task: write an OpenSCAD program to generate a Menger sponge.
[2025-05-27T15:28:45Z] Created menger_sponge.scad file with a recursive implementation that generates a Menger sponge fractal. The program allows for configurable size and iteration depth parameters.
[2025-05-27T15:28:50Z] WORK-COMPLETED
```
Installation
Installiere den Skill von https://github.com/pfcoperez/windsurfinaboxDieser Skill ist nicht in der ClawHub-Registry veröffentlicht. Füge den Prompt in deinen Coding-Agenten ein — Claude Code, Cursor, Codex — und er installiert den Skill direkt aus dem Quell-Repository.
Repository öffnen