For the complete documentation index, see llms.txt. This page is also available as Markdown.

Global flags

Flags available on every celigo command. These override config-file and env-var values.

Flag
Purpose

--token <value>

API bearer token. Overrides CELIGO_API_TOKEN and profile. Prefer the env var to avoid exposing the token in the process list.

--base-url <url>

API base URL. Overrides CELIGO_BASE_URL and profile.

--profile <name>

Use a specific profile for this command only.

--format <json|table>

Output format. Default json.

--jq <expr>

Transform JSON output through a jq expression.

--verbose

Print the underlying HTTP request and response (redacted).

-v, --version

Print CLI version and exit.

-h, --help

Show help for the current command.

--format

Default is json. Override per invocation, per profile, or via env var:

celigo flows list --format table       # one command
export CELIGO_FORMAT=table             # shell session
celigo config set default_format table # active profile (persistent)

Not every resource honors table — if a list command has no defined columns, it falls back to JSON regardless of --format.

--jq

Applies a jq expression to the JSON response. Useful for one-liners without piping:

--jq is incompatible with --format table — a jq filter implies JSON input and output.

--verbose

Prints each HTTP request and response to stderr. Auth headers are redacted.

Combine

All global flags can be mixed freely and placed before or after the subcommand:

Field projection on list commands

--fields and --limit are not global flags. They appear on list verbs, and they control how much data a listing returns. This section documents them here because the projection default affects almost every list command.

The default

A list command asks the API for a trimmed row rather than the complete resource document. Each row carries _id, name, and the fields the command's table columns need. This is the default because full collections are large: on one production account, celigo exports list returned 12.9 MB of JSON, and the projected listing returns 183 KB — 98.6% smaller. celigo flows list is 7.5 times smaller.

The lean shape is what most work needs. You list to find a resource, then get it by ID. get is never projected and always returns the complete document.

--fields

Value
Result

omitted

The command's built-in projection: _id, name, and the table columns.

default

The same built-in projection, stated explicitly.

all

Complete documents, as the API returns them.

a,b,c

_id, name, and the named fields. Dot notation selects nested paths, for example http.relativeURI.

An explicit field list is a floor, not an exact shape. _id and name are always requested, and so is any field the command reads off each row internally. celigo ai-agents list --fields lastModified requests lastModified and adaptorType, because the command filters on adaptorType client-side. Dropping that field would break the command, not just the display. Expect at least the fields you named — never fewer.

--limit

--limit <n> caps the number of rows and fetches a single page. It exists only on the standard list verb.

Seven listings are built differently and accept --fields without --limit:

  • ai-agents list

  • guardrails list

  • notifications list

  • stacks list

  • http-connectors list

  • trading-partner-connectors list

  • templates marketplace

To cap output from those, filter with --jq.

Precedence

Four inputs can decide the projection. They resolve in this order:

  1. --fields on the command

  2. --jq on the command, which requests complete documents

  3. The profile's list_fields setting — see Configuration reference

  4. The command's built-in default projection

--jq implies complete documents only when --fields is absent. A jq expression addresses the full document, so projecting underneath it would return null instead of an error. Combining --jq with an explicit --fields projects the rows first, and the expression then reads whatever the projection kept:

What is never projected

  • get on any resource, and every other single-resource read.

  • account snapshot, which builds the local index from complete documents.

  • jobs, sync-jobs, and recycle-bin listings.

  • storage list and storage search.

  • audit listings and the flow error listings.

Endpoints that ignore the request

The CLI requests a projection through the API's include query parameter. A few endpoints do not recognize it and return complete rows regardless. The CLI trims those rows locally, so the printed output matches the documented shape either way. Only the transfer size is unaffected.

Last updated

Was this helpful?