OpenAI Codex Multi OAuth

Manage and debug multiple OpenAI Codex OAuth profiles inside OpenClaw, including native multi-profile auth stores and external-router setups where one active...

This package appears coherent for debugging Codex profile selection, but be aware of the privacy/safety implications before running the scripts: they read your local OpenClaw auth store (which contains bearer tokens and profile metadata) and, if you run codex_usage_report.py, will use those tokens to call https://chatgpt.com/backend-api/wham/usage to fetch live usage. Recommendations: 1) Inspect the two Python scripts yourself (they are short and readable) before running. 2) Run summarize_codex_profiles.py first — it only reads local files and does not make network calls. 3) Back up ~/.openclaw/auth-profiles.json and any repo files before editing. 4) Only run codex_usage_report.py when you trust the endpoint (it is hard-coded to chatgpt.com) and you intend the script to use stored tokens to query usage. 5) If you want to avoid network usage, run the summarizer and manual inspection instead. 6) If you are unsure of provenance, run in an isolated environment or inspect file checksums / upstream source before trusting.

Type: OpenClaw Skill Name: openai-codex-multi-oauth Version: 1.2.1 The skill bundle 'openai-codex-multi-oauth' is designed for managing and debugging multiple OpenAI Codex OAuth profiles within the OpenClaw environment. It includes diagnostic scripts such as scripts/codex_usage_report.py and scripts/summarize_codex_profiles.py, which read local configuration files (e.g., auth-profiles.json) and session data to provide visibility into profile selection and usage. While the scripts handle sensitive OAuth tokens, they only communicate with the legitimate OpenAI backend (chatgpt.com) and do not exhibit signs of data exfiltration, malicious execution, or harmful prompt injection. The instructions in SKILL.md are consistent with the stated administrative and debugging purposes.