Use the CLI in automation

Use the Algolia CLI in scripts, CI, and agent workflows. Use structured output and DEBUG for troubleshooting, and preview changes before applying them.

This page explains how to use the Algolia CLI in automation, including non-interactive behavior, structured output, run summaries, and dry runs.

To learn more about interactive use, see Get started.

Credentials in automation

For environment variables, profiles, command-line flags, precedence, and Crawler keys, see Authenticate CLI commands.

Automation-focused choices:

• CI and other servers: Use ALGOLIA_APPLICATION_ID and ALGOLIA_API_KEY from your platform’s secret store so keys don’t end up in version control or config files.

• Local machines, including frequent CLI use by agents: Sign in once with algolia auth login, then select an application with algolia application select. The CLI stores credentials in your operating system’s keychain, so you don’t re-enter them for every command or leave them in shell history.

• One-off commands or temporary overrides: --application-id and --api-key can target a different Algolia application or API key for one command. On native Windows, these flags aren’t supported. Use environment variables instead. See Command-line options.

Non-interactive runs

In non-interactive environments, such as CI/CD, pipes, and agent workflows, the CLI doesn’t prompt for confirmation. Commands that delete or replace existing data require confirmation. To run these commands in a non-interactive environment, pass --confirm. Otherwise, the command exits with an error.

Structured output and command metadata

• JSON output. To keep standard output machine-readable, use -o json, -o jsonl, or -o ndjson where supported.

• Command schema. To help agents understand a command without parsing --help, run algolia describe <command>. This prints JSON metadata for the command, including flags and response fields.

Exit codes

The Algolia CLI uses the following conventions for exit code. Use them in scripts and CI to handle command outcomes programmatically.

Code	Meaning
0	Success
1	Error
2	User cancelled
4	Auth error

Pipe data into the CLI

To import data from another command or script, use -F - to read from standard input. Supported commands accept NDJSON or line-delimited JSON, depending on the command.

Command line

cat records.ndjson | algolia objects import INDEX_NAME -F -

Observability

Telemetry

By default the CLI sends usage analytics. Opt out with ALGOLIA_CLI_TELEMETRY=0. For details, see Telemetry.

Debug logging

Set DEBUG=1 (or any non-empty value) to print additional error details to stderr for telemetry and update notifier issues. For more information, see Telemetry.

Dry runs

To validate changes without applying them, use --dry-run. To get a structured preview, combine it with -o json.

Supported for these commands:

Command area	Examples
Application	algolia application create
Crawler	algolia crawler create, pause, reindex, run, test, unblock
Dictionary	algolia dictionary entries import
Indices	algolia indices delete
Objects	algolia objects update, operations, delete
Rules	algolia rules import
Synonyms	algolia synonyms import

Command line

algolia objects update INDEX_NAME -F records.ndjson --dry-run --output json

The output typically includes "dryRun": true and preview fields. The exact output depends on the command.

Use llms.txt for agent navigation

Algolia provides llms.txt files that help agents discover resources.

• Use algolia.com/llms.txt for a high-level map of Algolia, including product areas and key links.
• Use algolia.com/doc/llms.txt for documentation-focused navigation, such as API references and developer guides.

Environment variables (quick reference)

Variable	Effect
ALGOLIA_CLI_TELEMETRY=0	Opt out of telemetry
ALGOLIA_NO_UPDATE_NOTIFIER	Skip update checks when set to any value
DEBUG=1	Enable debug output on stderr when set to any non-empty value