Skip to main content
CDK Terrain v0.25.0 replaces the HashiCorp Checkpoint telemetry that the CLI inherited from CDKTF. The CLI no longer contacts checkpoint-api.hashicorp.com. It reports counted usage metrics to the CDKTN project’s own Sentry instead, and asks for your consent first (cdk-terrain#411, #412, #413). Usage metrics have their own flag in cdktf.json, sendUsageTelemetry. It’s separate from sendCrashReports, which controls crash reports:
  • CHECKPOINT_DISABLE keeps working as the environment-wide opt-out and takes precedence over cdktf.json.
  • The flags accept booleans or the strings "true" and "false". Any other value counts as missing.
  • The CLI only prompts when stdin and stdout are terminals, it isn’t running in CI, and the project has a cdktf.json to save the answer to.
  • cdktn init asks both questions for a new project. To answer them without a prompt, pass --enable-usage-telemetry=true|false and --enable-crash-reporting=true|false. Every built-in template writes both flags.
  • If you agree to usage metrics only, the CLI sends no crash reports.
The crash-reporting prompt now also appears in existing projects whose cdktf.json has no sendCrashReports key. Before v0.25 the CLI treated a missing key as “no” and never asked.

​
What the CLI reports

The metrics are counters and durations, aggregated per run. They cover:
  • Each command run: that it started, and whether it completed or failed, with the error category. Synthesis duration. A count of each CLI error type.
  • Environment: the command, the CI system, the project language, the operating system and architecture, whether the installed binary is Terraform or OpenTofu and its version, your declared targetVersions ranges, whether validateInstalledBinary is on, and the package manager.
  • Stacks: the backend type, counts of overrides, imports, and moves, the public registry providers you require (namespace/type) with their version constraints and whether the bindings are pre-built or generated, and the resource types you override. Counts are capped per run.
  • get and init: which providers and modules they generate, how many, and which built-in template init used.
  • convert: how many providers, modules, and lines it converted.
Sentry adds a random install ID, stored in ~/.cdktf/config.json. The CLI never sends stack names, construct IDs, resource addresses, file paths, your hostname or username, error messages, or SENTRY_* environment variables. Private registry hosts and module sources are reduced to a category such as private-registry, git, or local. These commands report metrics: convert, deploy, destroy, diff, get, init, list, output, synth, watch, provider add, provider list, and provider upgrade. login, debug, shell completion, --help, and --version don’t.

​
Learn more

Telemetry describes how to turn usage metrics and crash reports on or off, and Crash Reporting describes what a crash report contains. To set the flags, see Configuration File.