command_executed and command_failed) with three properties each (command, success, version).
Global flag capabilities
| Capability | What it gives you |
|---|---|
| Telemetry control | Disable telemetry per-run, per-environment, or persistently |
| AI/help formatting | Return help output in readable or structured formats |
| Cross-command consistency | Apply the same control flags to create, dev, start, preview, and build |
Global options
| Flag | Description | Default |
|---|---|---|
--no-telemetry | Disable anonymous telemetry for a single command execution. | telemetry enabled |
--ai-help | Show AI-assistant oriented help output. | disabled |
--format <pretty|json> | Output format used with --ai-help. | pretty |
Telemetry opt-out options
You can disable telemetry in three ways (listed in precedence order):EXTENSION_TELEMETRY_DISABLED=1 in continuous integration (CI) when policy requires no telemetry across all runs.
Practical examples
Disable telemetry for a CI build
Generate machine-readable help for tooling
Keep readable help output for local debugging
Behavior notes
- The CLI parses global flags before command-specific options.
- See repository
TELEMETRY.mdand Telemetry and privacy for telemetry behavior and privacy details. - Use
extension telemetry statusto check the current consent state. - Command pages document command-specific flags separately.
- Global flags can be combined with command-specific options.
- Machine consumers (AI/CI) should read command readiness from
dist/extension-js/<browser>/ready.json, not from terminal log parsing.

