Data collection

JetBrains Central CLI collects three types of analytics data:

Product usage events sent to Google Analytics 4 (GA4)
AI usage reported to JetBrains Central, where your organization can see it
Local request logs stored on disk

Product analytics (GA4)

JetBrains Central CLI sends anonymous usage events to Google Analytics 4 via the Measurement Protocol. This helps understand adoption, identify issues, and improve the product.

When product analytics are sent

Whether product analytics are sent depends on the choice stored in your configuration file:

User state	Value of `google_analytics_accepted`	Are product analytics sent?	What happens
Accepted analytics on the terms screen	`true`	Yes	Events from the interactive interface, the commands, and the proxy are sent
Declined analytics on the terms screen	`false`	No	No product analytics are sent
Has not seen or completed the terms screen	Not set	No	Nothing is sent until you make a choice on the terms screen
No configuration can be loaded	Not available	No	Product analytics are disabled

The interactive interface shows the terms screen before login on first launch, or before the main menu when consent is missing for an existing login. central login shows the same terms screen before the OAuth flow when consent has not yet been stored. Other non-interactive commands and the proxy do not show that screen; they use the stored choice if present, and send nothing if it is missing.

The stored choice can be changed at any time from the command line:

                central config set google-analytics on   # opt in
                central config set google-analytics off  # opt out
            

The current value is also shown in central config output — on, off, or not set.

What is tracked

Category	What is recorded
Onboarding	First launch, and each step of the login flow: started, browser opened, callback received, completed, or failed
Proxy	Proxy started, stopped, failed to start, or failed to restart
Tokens	Token refreshes and reloads, with how long each took and whether it succeeded
Requests	Per-request metadata: provider, response status, duration, and any retries
Agents	Agent configuration changes — connecting and disconnecting an agent
Interface and commands	Which menu action or command you ran: viewing your usage, viewing logs, changing a setting, switching AI access source, checking for updates, and uninstalling. The interactive interface and the equivalent command are recorded separately, so the two can be told apart.
Updates	Whether an update was installed or failed, and whether you started it yourself or an agent installed it when a session began
Data migration	Whether a configuration or data-directory migration ran, and whether it had to be rolled back

What is not tracked

Product analytics events never include:

Request or response bodies — no prompts, no completions
Plaintext email addresses
File contents or code
Specific error messages from LLM responses
API keys or tokens

User identification

The user ID is a truncated SHA-256 hash of the email address, 16 hex characters. No plaintext email is ever sent:

                email:   user@jetbrains.com
                user_id: SHA256("user@jetbrains.com")[:16] → "a1b2c3d4e5f6a7b8"
            

Analytics session identification

Events are grouped into sessions by a random identifier. Events from one proxy run share a session.

This identifier groups product analytics events only. It is not the AI session described in AI usage reported to JetBrains Central.

Event context

Every event includes:

JetBrains Central CLI version
Operating system and architecture
Timestamp
Release track — the resolved update channel (stable or beta), or pinned when pinned to a specific version. The pinned version itself is not sent.

Delivery

Events are batched and sent periodically, or immediately when the proxy stops. Failed sends are retried a few times and then discarded. Analytics failures never affect how JetBrains Central CLI works.

AI usage reported to JetBrains Central

Agents you route through JetBrains Central CLI report their AI usage to JetBrains Central as sessions. A session is a single run of an agent. This is separate from the product analytics above: it is sent to JetBrains Central rather than to Google Analytics, and the people who can see it are the administrators of your own organization.

What each session records

A session identifier
The user or service account that started the session
When the session started, and the time of its last activity
How long the session ran
The agent that ran the session, and the models it used
How many tokens the session consumed
How many AI credits the session consumed

Where this data is visible

Administrators of your organization see this data in the AI analytics section of JetBrains Central Console, and can also read it through the Analytics API:

Session explorer lists individual sessions.
AI analytics aggregates AI credit consumption and token usage across the organization.

Reporting delay

Session data is collected in batches, so it does not appear immediately. For a long-running session, the duration, token count, and AI credit count can still change after the session first appears.

Local request logs

All proxy requests are logged to ~/.jetbrains-central/analytics/ for local debugging. These logs stay on your machine.

Logged fields

Field	Description
`timestamp`	When the request was made
`original_url`	The agent's request URL, with the proxy secret masked — for example, `/wire/***/claude-code/anthropic/v1/messages`
`final_url`	Cleaned platform path, with no host or credentials
`method`	HTTP method
`status`	Response status code
`duration_ms`	Request duration
`is_error`	Whether the status is 400 or above
`error_message`	Error details, if applicable
`grazie_trace_id`	Trace ID you can quote in a support request
`ingrazzio_trace_id`	Additional trace ID, present only for Junie requests on a JetBrains Account login

Viewing logs

In the JetBrains Central CLI interactive interface, select More... | AI Logs to browse request logs:

t cycles the time filter — all, last hour, last day
s cycles the status filter — all, success, errors
r resets all filters
R refreshes the data
The arrow keys page through the history
q goes back to the menu

Failed requests show the error message and a trace ID you can quote in a support request.

Log retention

At most 1000 entries per log file
At most five log files, with the oldest deleted automatically
Files are named ~/.jetbrains-central/analytics/requests_{timestamp}.json

Clearing logs

rm -rf ~/.jetbrains-central/analytics

The directory is recreated automatically when the proxy logs the next request.

29 September 2026