Telemetry
Hotspots has opt-in, anonymous usage telemetry. It is off unless you turn it on.
What I'm asking for
I'm a solo maintainer and I can't see how hotspots is used. Download counts include CI runs and mirrors, so they don't tell me how many people actually run it. If you're willing, enabling telemetry sends one small anonymous ping a day (the fields below, nothing else). It helps me answer a few questions:
- How many real installs are active, and does anyone keep using it after the first run?
- Which versions are still in use, so I know when it's safe to drop old behavior?
- Which install channels (
npm,pypi,cargo,brew, binary) and platforms to test and support first? - Which subcommands get used, so I improve those rather than guess?
That's all I use it for. I don't sell it, share it, or join it to anything else. You can say no, and the tool works exactly the same.
What is sent
At most one successful ping per 24 hours, only while enabled. If a ping fails (offline, firewalled, server down) it is silently retried after about an hour, never more often:
{
"name": "cli_ping",
"meta": {
"v": "2.5.0",
"os": "linux",
"arch": "x86_64",
"channel": "npm",
"cmd": "analyze",
"iid": "9f2c4e1a7b3d58f0a1c6e2d4b7f90312"
}
}| Field | Meaning |
|---|---|
v | hotspots version |
os | Operating system (e.g. linux, macos, windows) |
arch | CPU architecture (e.g. x86_64, aarch64) |
channel | Install channel guessed from the executable path: npm, pypi, cargo, brew or binary |
cmd | Name of the subcommand that ran, never its arguments |
iid | Random install id, generated the first time a ping is sent while telemetry is enabled. Not derived from your machine, user or paths |
That table is the complete list. It mirrors COLLECTED in hotspots-cli/src/cmd/telemetry.rs, which hotspots analyze --show-config prints a summary of, and a test fails if the code and this table disagree.
Added by the server on receipt: the time of the request, the request's country (from the CDN; your IP address is not stored) and the fixed User-Agent hotspots-cli. Nothing else is stored.
Never collected: source code, file or repo names, paths, languages, scores, analysis results, command arguments, environment variables, hostnames, usernames, IP addresses.
Controls
There is no telemetry subcommand. Telemetry is controlled entirely by one environment variable:
export HOTSPOTS_TELEMETRY=on # enable (persists for the shell session, or add it to your profile)
unset HOTSPOTS_TELEMETRY # disable — the default; any value other than "on" also disables ithotspots analyze --show-config # shows a line with the current telemetry state and what it sendsIt is off by default, with no interactive prompt — set HOTSPOTS_TELEMETRY=on explicitly to turn it on, in a shell profile or CI environment config, rather than waiting to be asked.
Environment overrides that always win over HOTSPOTS_TELEMETRY=on:
DO_NOT_TRACK=1- any of
CI,GITHUB_ACTIONS,GITLAB_CI,BUILDKITE,JENKINS_URL
So setting HOTSPOTS_TELEMETRY=on globally (e.g. in ~/.bashrc) never causes a CI runner to ping, even if that CI environment happens to inherit the same shell profile.
State lives in ~/.hotspots/telemetry.json (your random install id, and when a ping last succeeded or failed); deleting it resets the install id. The request is cut off after about 3 seconds (half a second if the network is unreachable), failures are silent and never change output or exit codes.
Why
To learn how many installs are actually in use and which versions and install channels matter, since download counts include CI and mirrors.