wflow CLI

The wflow.com public API from the command line. Every API operation is one command, every command prints JSON (except a download to --output -, which writes the file's raw bytes to stdout), and wflow <command> --help documents what it sends and returns.

Install

pipx install wflow-cli        # or: uv tool install wflow-cli   or: pip install wflow-cli
wflow --version

Python 3.12 or newer. pipx and uv tool keep the CLI in its own environment.

Set up an API client

The CLI signs in as an API client with the client_credentials grant — a machine identity, not a person.

1. Create an API client with the client_credentials grant in admin.wflow.com.
2. Allow the client in each workspace it should reach: in the workspace, Settings → Integrations → API accesses.
3. Give the CLI the credentials:

export WFLOW_CLIENT_ID="your-full-client-id"
export WFLOW_CLIENT_SECRET="your-client-secret"
export WFLOW_ORGANIZATION="your-workspace"   # optional: the default for --organization/-o

Credentials are read from the environment only; no option takes a secret.

Use it

wflow --help                                              # every command, grouped by area
wflow get-organization -o acme
wflow list-register-partners -o acme --page-size 50 --query 'name = "Acme s.r.o."'
wflow list-documents -o acme --sort "created desc"

A list command that takes --page and --page-size returns one page: {items, totalItems, pageSize, page}. Compare totalItems with what you have before assuming you have everything.

Writing

Options that take a record or a list of records take JSON — inline, from a file with @, or from stdin with - (one option per command can read stdin):

wflow patch-register-partners -o acme --partners @partners.json
cat partners.json | wflow patch-register-partners -o acme --partners -

update-register-* replaces the whole register: every record you leave out is retired. patch-register-* leaves the records you do not send alone. Both replace each record you send as a whole, so send the full record. wflow keeps no version history — read wflow <command> --help before writing.

Files

wflow upload-document-file -o acme --document-id <id> --file ./invoice.pdf
wflow download-document-file -o acme --file-id <id> --output ./invoice.pdf
wflow download-document-file -o acme --file-id <id> --output - > invoice.pdf

An upload is stored under the file's own name unless you pass --filename. A download never overwrites an existing file unless you pass --force.

Exit codes

For AI agents

Every command prints its result as JSON on stdout and the reason for a failure on stderr, so an agent can run wflow from a shell, check the exit code and parse the result. The one exception is a download with --output -, which writes the file's raw bytes to stdout instead. wflow <command> --help is the documentation for that command.

Python

The commands are generated from the wflow-api package, which you can use directly.