Joshua Shay Kricheli NeuroSymbolic AI
v2 v1

Release · v2.26.6

Repair deploy-vercel run-name so the workflow…

fix(ci): repair deploy-vercel run-name so the workflow validates

fix v2.26.6 June 25, 2026 Claude 5728fc8

Details

The first run came back "No jobs were run" / failure. Root cause: the
`run-name` value was unquoted, so the ` #` inside its `format('PR #{0}', …)`
expression was parsed by YAML as a comment, truncating the `${{ }}` expression.
GitHub's workflow validator then rejected the file and produced an empty,
failed run (lenient parsers like PyYAML silently truncated instead, so it
looked fine locally). Caught with actionlint.
- Quote the run-name value so the `#` is literal.
- Make the gate the first, always-running step (drop the job-level `if`) so the
  job is never empty even if a future trigger fires unexpectedly; the gate now
  also skips fork PRs (no secret access) in addition to the missing-token case.
- Move the production/preview check to step scope (off the workflow-level env).
Validated with actionlint (clean).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016HyukoK8KXyZw4hY9Wm3zN

Files changed (1)

.github/workflows/deploy-vercel.yml +24 −15