Codex Desktop hides automation when automation.toml is UTF-8 with BOM

Open 💬 1 comment Opened Mar 8, 2026 by cfvbaibai

Summary

In Codex Desktop, an automation can disappear from the UI when its automation.toml file is saved as UTF-8 with BOM.

This is a different surface from SKILL.md parsing: this affects automation discovery/loading in Desktop.

Environment

  • Codex Desktop (Windows)
  • Automation stored at: ~/.codex/automations/<id>/automation.toml

Reproduction

  1. Create or edit an automation TOML under ~/.codex/automations/<id>/automation.toml.
  2. Save file as UTF-8 with BOM (file header starts with bytes EF BB BF).
  3. Open/refresh Codex Desktop automation list.
  4. Observe the automation is missing/disappears.

Observed behavior

  • Automation is not shown in Codex Desktop.
  • File content is otherwise valid TOML.

Expected behavior

  • Desktop should load TOML regardless of UTF-8 BOM, or show explicit parse error in UI/logs.

Workaround

  • Re-save automation.toml as UTF-8 without BOM.
  • After rewrite, file header starts with text bytes (e.g. 76 65 72 73 for vers), and automation reappears.

Notes

  • We confirmed this repeatedly in real use.
  • Please normalize BOM handling for automation TOML parser path.

View original on GitHub ↗

This issue has 1 comment on GitHub. Read the full discussion on GitHub ↗