Skip to content

Quick start

This page runs the same job, compressing a photo, in each interface. Pick the section you need.

  1. Open Dogubako (the desktop app, or your server in a browser).
  2. Drag photo.jpg anywhere onto the window, or click Choose files.
  3. Dogubako asks What should happen to it? and numbers the likeliest tools. Press 1 for the top one, or type a word such as smaller and press Enter.
  4. On the tool page, adjust the settings if you like and run the tool. The original and the result appear side by side with the size change.
  5. Download or save the result. It also lands on your bench on the Home page, ready for the next tool.

Every tool is also one search away: press ⌘ K (or Ctrl K) and type its name. The app tour explains every screen.

Terminal window
dogubako image compress photo.jpg # writes photo-compressed.jpg in the current folder
dogubako image compress photo.jpg -o small.jpg
dogubako image compress --help # every option, generated from the tool

Find tools by keyword and inspect one:

Terminal window
dogubako tools search compress
dogubako tools info image.compress

More in the CLI overview and recipes.

Against a self-hosted server, with an API token:

Terminal window
H="Authorization: Bearer $DOGUBAKO_TOKEN"
HOST=http://localhost:8420
# 1. upload the file and keep its id
ID=$(curl -s -H "$H" -T photo.jpg "$HOST/api/v1/files/raw?name=photo.jpg" | jq -r .id)
# 2. run the tool, wait up to 60 seconds, and keep the output's id
OUT=$(curl -s -H "$H" -H 'Content-Type: application/json' \
-d "{\"params\":{\"input\":[\"$ID\"]},\"wait_seconds\":60}" \
"$HOST/api/v1/tools/image.compress/run" | jq -r '.result.artifacts[0].id')
# 3. download the output
curl -H "$H" -o small.jpg "$HOST/api/v1/files/$OUT/content"

The HTTP API reference covers jobs, progress events and errors.

For an agent on the same machine, register the stdio MCP server, here with Claude Code:

Terminal window
claude mcp add dogubako -- dogubako mcp serve --discovery compact

Then ask the agent to “compress photo.jpg with Dogubako”. For a shared server with a token per agent, see Remote agents.