Deployment

A deployment is solver code under run/ and its dataset under data/, published under a name your company owns. aura deploy creates one from the config table describing it; every other verb here acts on one that already exists.

aura deploy

aura deploy --dry-run     # print the resolved file list, upload nothing
aura deploy               # upload, then run the server-side runnability check

Option	Does
[name]	Which table to read, when the file defines more than one
--config <path>	The pyproject.toml to read (default: the nearest one with [tool.aura])
--dry-run	Print what would upload, change nothing
--verbose	Log every file — + new, ↻ replaced — instead of one summary line per run

The uploads run, then the runnability check; a failing check exits non-zero. On success it prints the aura project create command for what you just deployed.

Redeploying is a sync

Remote run/ and data/ are made to match your local folders: after the uploads, remote files no longer present locally are removed, so renames and deletions leave nothing stale. A failed upload stops the sync before it prunes.

Agent config syncs only where you declared it. Omitting agents_md or skills_folder leaves whatever is already uploaded alone; declaring one makes the local tree authoritative, and remote entries missing locally are then pruned.

Commands

Command	Does
aura deployment list	List your company’s deployments
aura deployment check [name]	Run the server-side runnability check
aura deployment delete <handle|name>	Delete a deployment and every file in it (y/N prompt; --yes to skip)
aura deployment file list [name]	Tree of deployment files (--prefix <path>, --flat)
aura deployment file upload <local> <remote> [name]	Upload one file, replacing whatever is at that path
aura deployment file delete <remote> [name]	Remove one file (y/N prompt; --yes to skip)
aura deployment instructions show [name]	Print the instructions it ships
aura deployment instructions set <file> [name]	Replace them with a file’s contents
aura deployment instructions delete [name]	Remove them
aura deployment skills list [name]	List its skills, with descriptions
aura deployment skills show <skill> [name]	Print one skill in full, file contents included
aura deployment skills set <dir> [name]	Upload one skill from a directory
aura deployment skills delete <skill> [name]	Remove one skill

Deleting a deployment is permanent

aura deployment delete purges every file under it and frees the name, with no undelete. Projects already created from it keep their own copy and go on working.

Changing a deployment’s instructions or skills sets what future projects inherit. It never reaches a project that already exists — that project holds its own copy, and pulls a revision in only when you ask it to.

Changing one file

aura deploy is a sync, so it prunes. To swap in one data file or patch one solver module, use aura deployment file, which touches nothing but the path you name.

aura deployment file list --prefix run/src/                  # what is up there
aura deployment file upload ./src/solver/model.py run/src/solver/model.py
aura deployment file upload ./tuesday.csv data/orders.csv    # rename on the way up
aura deployment file delete data/stale.csv --yes

• The remote path is always explicit. A deployment’s layout is not your local one — aura deploy puts model_folder under run/ and data_folder under data/ — so nothing is inferred from the local path. One file per command.
• Uploads must land under run/ or data/. Nothing else is part of the run contract. file delete takes any path file list prints, since removing a file cannot create a stray one.
• The content type comes from the remote path, which is the name a run reads — so renaming on the way up types the file by its destination.
• The runnability check runs after both. Pass --no-check to skip it, for the middle of a && chain.

An upload replaces without asking, and says which it did

There is no versioning and no undo, so each upload marks which of the two things happened:

+ run/README.md    28 bytes (new)
↻ data/orders.csv  18 → 28 bytes (replaced)

+ added a file; ↻ overwrote one, and those bytes are gone. A + you did not expect means the remote path is wrong and a run is still reading the old file — a mistake the runnability check cannot catch, because a deployment with one stray extra file is still perfectly runnable.

aura deploy --verbose marks its per-file log the same way, and its summary counts them.

Naming a deployment

[name] in brackets is optional: omit it and the command reads the name from the nearest config table, or from --config <path>. It never means “every deployment” — each command acts on exactly one, and outside a model folder it fails rather than guessing.

cd my-model/solver
aura deployment skills list        # the deployment this folder defines
aura deployment skills list vrp    # named outright, from anywhere

Either spelling works: vrp, or the full acme/vrp handle you get from aura project create. A handle whose company is not yours is refused rather than resolving to your own deployment of that name.

The API itself takes the bare name

/api/v1/deployments/{name}/… accepts only the bare name — an encoded / is decoded before routing, so a handle 404s at the router. The CLI splits it for you and checks the company; the SDK methods pass the name straight through. Where the API does accept a handle it arrives in a request body, as with POST /projects/{id}/inherit.