v0.1.4 rebuilds codex-grok-search around one first principle: deliver Grok results reliably.
Highlights:
- Reduces the wrapper to a thin Codex-to-Grok bridge running outside the current repository with a temporary isolated home.
- Removes platform, URL, date, field, Markdown, version, login, model-catalog, and inspect preflight gates that could discard successful searches.
- Removes local Reddit refetching and treats platform selection as a search hint instead of an exclusion rule.
- Saves Grok’s free-form answer verbatim and returns usable output as
partialeven if the process later exits non-zero. - Keeps seven-day result retention and
--keep-run, while removing the maximum-run capacity blocker. - Keeps quick mode free of automatic Codex browser verification unless the user explicitly requests it.
Validation:
- 28 unit tests passed locally on Python 3.9.6.
- GitHub Actions passed on Python 3.9 and 3.12: https://github.com/sudoHG/codex-grok-search/actions/runs/29989752441
- A real Grok 4.5 X-search canary completed from the isolated run directory without browser verification.
- Release assets were rebuilt twice from commit
d3bf1e2c30495658c6cc08f8ce440b9fa1e62ad7with Python 3.9.6 and matched byte for byte. - ZIP SHA-256:
b7aca555bf1e8ce142ad2053f32c5299e04eff2b9103010efd1ea8f8d672aa6b - tar.gz SHA-256:
da6e7926b276f82e3cc9b858754e475a3e11210dc9603de6996405663f8d87c1