Summary

safe-outputs operations using target: \"triggering\" can incorrectly fail in pull_request_target runs with:

Target is \"triggering\" but not running in pull request context

Repro

• Trigger workflow on pull_request_target (e.g. synchronize).
• Use reusable gh-aw workflow with a PR-scoped safe output (example: update-pull-request with default target: \"triggering\").
• Observe warning/error in safe outputs job:
  • https://github.com/elastic/ai-github-actions-playground/actions/runs/22840740422/job/66246465939

Expected

pull_request_target should be treated as PR context for target: \"triggering\" resolution, same as pull_request.

Actual

Resolver rejects the context and skips/fails the message as not in PR context.

Suspected cause

actions/setup/js/safe_output_helpers.cjs PR-context detection for resolveTarget() includes:

• pull_request
• pull_request_review
• pull_request_review_comment

but excludes pull_request_target.

Proposed fix

Include pull_request_target in PR-context detection in resolveTarget() and add regression tests in safe_output_helpers.test.cjs.