fuse init

Scaffold an example Fusefile

fuse init [path] [flags]

Writes a fully commented example Fusefile to disk, documenting the common v1 fields with an explanatory comment on each. The GPU fields under resources (gpu, gpu_kind, gpu_profile) are not part of the scaffold, see Fusefile for those.

It also writes a .fuseignore next to it, which bounds what a copy: block ships. Every line in it is a comment: the default ignores (.git/, node_modules/, .env and friends) apply with or without the file, so the scaffold exists to show you what they are and where to override one, not to change anything.

The scaffold is deliberately larger than a working file. It names every field so you can see what exists, which also means it is not runnable as written: it declares a secret you have to supply at up time, and an image that must already be baked on the host. If you want the opposite, the fewest lines that boot with a bare fuse up, see Your first Fusefile.

Flags

Examples

fuse init
# writes ./Fusefile and ./.fuseignore

fuse init ./services/api/Fusefile
# the positional argument is the file to write, not a directory

fuse init --force
# overwrites an existing ./Fusefile and ./.fuseignore

Without --force, init refuses to overwrite an existing file at the target path, protecting any edits you’ve already made. Passing a directory path is not a shortcut for the Fusefile inside it, it aborts with a Fusefile already exists at ./services/api (use --force to overwrite) because the directory itself exists.

An existing .fuseignore stops it the same way, naming that file instead. Both are checked before either is written, so a refusal never leaves you half scaffolded.

After scaffolding, edit the file to match your workload, then run fuse up to bring it up.