AI agents removed the friction from writing telemetry
I used to avoid adding telemetry because it felt like tedious busywork. Now with Claude handling the OpenTelemetry boilerplate, I'm instrumenting everything.
Before tagging a new version, lerna scans your codebase for changed files. If it finds a changed file, it'll recommend a new version. This is really useful most of the time, but you only want to bump to a new version when some library functionality changes. You certainly don't want to introduce a new version just because you changed the README.md file.
There are two ways that you can tell lerna to ignore specific changed files.
You can use the --ignore-changes flag to prevent lerna from publishing new versions when it finds changes in specific file types.
If you'd prefer, you can add an ignore key in the lerna.json file.
I used to avoid adding telemetry because it felt like tedious busywork. Now with Claude handling the OpenTelemetry boilerplate, I'm instrumenting everything.
AI agents work better when given appropriate context and guardrails.
I've been using Claude Code to offload tedious parts of platform engineering: dependency reviews, generating test PRs, dependency migrations, and project cleanup.