I’m writing some utility code for custom GitHub actions. It needs to find its own Step name as it appears in the Web UI. How can this be done?

What I’ve tried so far.

• In contrast to the GITHUB_WORKFLOW and GITHUB_JOB env vars, the GITHUB_ACTION env var does not contain the name. The former contain the names as they appear in the yaml, the latter only contains the id of the step, and if there is no id, a generated name, like __run, or __actions/checkout_2. It is also not present in any context object according to the GitHub docs.
• The name seems to be available in the API. However, to use the name from the API, i need to find the correct object in the json that represents the entire workflow run. The GITHUB_RUN_ID makes it easy to identify the workflow. Identifying the job is a bigger problem though. There is no GITHUB_JOB_ID that would make it easy to identify. Using the name is impossible, because the GITHUB_JOB env var contains the name as it appears in the yaml, however, the name from the API contains both a prefix in case its a re-usable workflow representing the calling workflow as well as a postfix in case its a matrix execution. Assuming we would have overcome this, identifying the correct step in the job is just as hard. In theory, since steps are only executed sequentially, the “current” step should be the only one with the state in_progress. Unfortunately, the API seems to be lagging behind reality, by up to 60 seconds, i.e., the step with the state in_progress is never the current step. Just in case, the lag is introduced by a HTTP Proxy / server-side cache, I tried adding a Cache-Control: no-cache header, but without success.

How can this be done?