Global Options and Auth

Global options are available for all Flexkit CLI commands.

Global Options

Option	Description
`-h`, `--help`	Output usage information.
`-v`, `--version`	Output the CLI version.
`--cwd <DIR>`	Set the working directory for this command invocation.
`-A`, `--local-config <FILE>`	Path to a local `flexkit.config.[js\|ts]` project config file. `.tsx` is not accepted.
`-Q`, `--global-config <DIR>`	Path to the global configuration directory.
`-d`, `--debug`	Enable debug output, including every API request and its response status.
`--no-color`	Disable colored output.
`-S`, `--scope <SCOPE>`	Set a custom scope (used by commands that support scoped operations, such as `project`).
`-t`, `--token <TOKEN>`	Use a token for this run instead of persisted credentials.
`--api <URL>`	Send requests to another API host than `https://api.flexkit.io`.

Options shared by project commands

Commands that act on a project’s data (automations, records, skills, query and the others) also accept these options:

Option	Description
`-p`, `--project <ID>`	The project to act on. See how commands choose a project.
`--json`	Print machine-readable JSON on stdout instead of tables.
`-y`, `--yes`	Skip confirmation prompts for destructive actions, such as `rm`.

Choosing a project

Commands resolve the project they act on in this order:

The --project option.
The FLEXKIT_PROJECT_ID environment variable.
The nearest .flexkit/project.json, written by flexkit link, searched from the current folder upwards.
The project in the local flexkit.config.ts, when it defines exactly one.

Environment variables

Variable	Description
`FLEXKIT_TOKEN`	API token to use instead of the stored login. `--token` takes precedence.
`FLEXKIT_PROJECT_ID`	Project to act on when `--project` is not given.
`FLEXKIT_API_URL`	API host to use, like `--api`.

Authentication Behavior

Commands that require auth prompt for login in TTY mode if no credentials exist.
In non-interactive environments, set FLEXKIT_TOKEN or pass --token. Create tokens in the dashboard; see API tokens.
Token format must be flk- followed by 32 hexadecimal characters.
login, logout, whoami, unlink and help can run without an existing token.
Every command acts with your own role and space memberships. Commands that need a person (for example chat) say so on their page; with a token they fail with a clear error.

Output and exit codes

Tables, progress and messages are written to stderr. Data requested with --json is written to stdout, so flexkit automations ls --json | jq always receives clean JSON.
Commands exit with 0 on success, 1 on errors, and 2 after printing help.
Destructive commands ask for confirmation. Without a terminal they refuse to run unless you pass --yes, so a script never hangs on a prompt.

Common Examples

# Show CLI version
flexkit --version
 
# Use a specific project config file
flexkit --local-config ./config/flexkit.config.ts deploy
 
# Run in CI with a token and a project from the environment
FLEXKIT_TOKEN=flk-... FLEXKIT_PROJECT_ID=abcdefghij flexkit automations ls --json
 
# Act on another project than the linked one for a single command
flexkit records ls product --project abcdefghij
 
# Delete without a prompt in a script
flexkit skills rm skl_123 --yes
 
# See every API request a command makes
flexkit automations ls --debug