OpenAI Auth Switcher Public

Use this skill as the publishable public-track release of the OpenAI auth switcher workflow.

It is designed for OpenClaw administrators who want a web-first, first-run-friendly, release-safe workflow for OpenAI OAuth account takeover, inspection, dry-run validation, controlled switching, and public distribution.

Purpose

Keep the live/internal operator skill and the public distributable skill separated.

This public track must:

• - avoid bundling live runtime state
• avoid bundling auth snapshots, callbacks, backups, or token ledgers from a real machine
• avoid machine-specific hard-coded paths where possible
• document compatibility boundaries explicitly
• provide a release-safe packaging workflow
• keep temporary runtime files inside the skill runtime area, while encouraging important persistent state to live in an external state base via INLINECODE0

Core operating model

Treat OpenClaw OpenAI OAuth switching as a high-sensitivity maintenance workflow.

Always do work in this order:

1. 1. Use install.sh as the default user-facing bootstrap entrypoint.
2. Run doctor.py when installation or environment checks fail.
3. Confirm runtime discovery with env_detect.py.
4. Inspect the current runtime before any switch logic.
5. Dry-run any target before proposing a write.
6. Keep rollback and backup behavior explicit.
7. Package only from this public skill directory or from a sanitized staging copy.

Included scripts

Primary public-release scripts:

• - install.sh — recommended user entrypoint; wraps the web bootstrap into a single shell command
• INLINECODE5 — recommended cleanup entrypoint before INLINECODE6
• INLINECODE7 — one-shot web bootstrap for first-run access
• INLINECODE8 — port selection helper (9527 → 12138 → fallback)
• INLINECODE11 — default admin credential generator
• INLINECODE12 — compatibility and environment checks
• INLINECODE13 — OpenClaw path and runtime discovery
• INLINECODE14 — centralized path resolution helpers
• INLINECODE15 — portable runtime inspection
• INLINECODE16 — public-safe slot metadata and local slot files
• INLINECODE17 — rollback helper using explicit backup sources
• INLINECODE18 — controlled switch experiment with backup and rollback
• INLINECODE19 — local token attribution ledger rebuild
• INLINECODE20 — hourly/daily rollup payload for local analytics
• INLINECODE21 — release-safe packager wrapper

Helper modules:

• - INLINECODE22
• INLINECODE23
• INLINECODE24
• INLINECODE25

Compatibility and safety references

Read only as needed:

• - references/compatibility.md — Python / Node / OpenClaw / OS expectations
• INLINECODE27 — path detection and override model
• INLINECODE28 — operator flow and first-run checks
• INLINECODE29 — sensitivity, boundaries, and redaction rules
• INLINECODE30 — publish checklist and forbidden contents
• INLINECODE31 — relation to the internal/live skill

Release rule

Do not publish skills/openai-auth-switcher directly.

Use the public skill directory for ClawHub publication and use a packaging wrapper that rejects runtime data, backups, session callbacks, and credential-bearing files.

Recommended first release positioning:

• - version: INLINECODE33
• tested on OpenClaw INLINECODE34
• tested on Python INLINECODE35
• tested on Node.js INLINECODE36
• Linux-first release