Telemetry

The anonymous usage data the Queria CLI collects and how to opt out

The Queria CLI / MCP server collects anonymous usage data to improve the tool. You can opt out at any time.

This page covers two separate mechanisms.

	Usage data (first half)	Feedback (second half)
What is sent	Fixed fields such as the command name	Text written by you or an agent
When it is sent	Automatically (opt-out)	Only when you ask for it (opt-in)
How to stop it	`queria telemetry disable`	Nothing is sent by default

Usage data never contains free text. Feedback is something you send yourself, when you want to tell us about a request or a defect.

What is collected

One event is sent per command (or MCP tool call):

Command name (list / search / sql and so on; the tool name for MCP)
Success or failure
Package version
Frontend kind (cli / mcp)
Target dataset name (only when given as a command argument)
Whether it ran under an agent (detected from known Claude Code / Codex environment variables)
An anonymous client ID (randomly generated on first run and saved to ~/.config/queria/config.toml)

What is not collected

SQL query text
File paths and output contents
Personal information such as names or email addresses
IP addresses (never stored)

Where events go

Events are sent only to Queria’s own endpoint, telemetry.queria.io. Nothing is sent to third-party analytics services.

When an API token is registered, usage is associated with that account. queria logout (or queria auth clear) stops the association. If you pass the token through the QUERIA_TOKEN environment variable, unset that variable as well.

Opting out

Disable with any of the following:

queria telemetry disable        # saved to the config file (telemetry = false)
queria telemetry status         # check the current state

Environment variables also work:

export DO_NOT_TRACK=1           # the standard variable (consoledonottrack.com)
export QUERIA_NO_TELEMETRY=1

To re-enable, run queria telemetry enable.

Sending feedback

Reports a dataset request, a data defect, or a documentation error to Queria. Nothing is sent unless you ask for it.

queria feedback send "I need childcare waiting-list counts by municipality" --kind dataset_request

Before sending, the exact payload is printed and you are asked to confirm.

This sends the text below to Queria so we can act on it. Do not include
credentials, personal data or query results.
Privacy policy: https://queria.io/legal/privacy-policy
To: https://telemetry.queria.io/feedback
  agent: claude-code
  app_version: 0.15.0
  client_id: 81b4805ebe12423ebd1494c590ca89a3
  kind: dataset_request
  submitted_via: human
  message: |
    I need childcare waiting-list counts by municipality
Send this? [Y/n]

--kind says what sort of report it is.

Value	Use for
`dataset_request`	The data is not in the catalog at all
`data_issue`	The data exists but is wrong, stale, or has gaps
`docs_issue`	The documentation is wrong, missing something, or unclear
`cli_issue`	The CLI or MCP server itself misbehaves
`other`	Anything else, including plain thanks (default)

Add --dataset <name> when the report is about a specific dataset. It is what lets us count how many separate machines asked for the same thing, which is how requests get prioritised.

--yes skips the confirmation prompt. It does not skip printing the payload. Outside a terminal (no TTY), running without --yes exits without sending, because the prompt cannot be answered.

What is sent

The message (up to 4000 characters; it is sent as written, so anything longer is refused rather than truncated)
The kind, and the dataset name
The anonymous client id (the same one as usage data), the tool version, the frontend (cli / mcp), and the agent runtime name

There is no contact field. If you need a reply and have no API token configured, say so in the message.

Letting an AI agent send it

An agent needs explicit prior permission, per machine, before it can send on your behalf. Without it, an agent cannot send the text at all.

queria feedback status         # show the current setting
queria feedback allow         # allow (each message still needs your approval)
queria feedback allow --auto  # allow, and skip the per-message approval
queria feedback disallow      # revoke

Instead of typing a command, you can just answer the agent’s question — it records your answer with set_feedback_consent.

allow (recommended) — the agent must show you the full text and get your approval. The first submit_feedback call only returns a draft; it does not send
allow --auto — the agent may send without asking each time. Anything sent this way is recorded as a report no human reviewed

The setting lives in ~/.config/queria/config.toml and applies to every agent on that machine.

How submissions are handled is set out in article 6 of the privacy policy. To have something you sent deleted, contact us through that policy (queria feedback status shows your client id).