PR #63447: feat(openai-codex): add gpt-5.4-pro forward-compat #63404

• Repository: openclaw/openclaw
• Author: jepson-liu
• State: open | merged: False
• Link: https://github.com/openclaw/openclaw/pull/63447

Description (problem / solution / changelog)

Summary

• Problem: ChatGPT Pro users selecting openai-codex/gpt-5.4-pro hit model not allowed: openai-codex/gpt-5.4-pro before any API call because the Codex provider did not forward-compat resolve or advertise that model (see issue #63404).
• Why it matters: OAuth Codex users on Pro plans expect the same model id family as the API-key openai path; blocking gpt-5.4-pro breaks a primary workflow.
• What changed:
  • extensions/openai/openai-codex-provider.ts: add gpt-5.4-pro to forward-compat resolution, XHigh/modern id lists, synthetic catalog entry, Pro pricing metadata, and preferRuntimeResolvedModel for both gpt-5.4 and gpt-5.4-pro.
  • Tests/harnesses aligned so mocks and Pi test support match real plugin behavior (model compat, thinking levels, Discord /think autocomplete mock, models list forward-compat, provider-runtime catalog expectations, e2e harness).
• What did NOT change: Default Codex default model remains openai-codex/gpt-5.4. No change to Issue’s optional follow-up (minimum thinking tier for gpt-5.4-pro); can be tracked separately if the upstream requires non-none thinking.

Change Type (select all)

• Bug fix
• Feature
• Refactor required for the fix
• Docs
• Security hardening
• Chore/infra

Scope (select all touched areas)

• Gateway / orchestration
• Skills / tool execution
• Auth / tokens
• Memory / storage
• Integrations
• API / contracts
• UI / DX
• CI/CD / infra

Linked Issue/PR

Closes #63404

User-visible / Behavior Changes

• Users can select and run openai-codex/gpt-5.4-pro when authenticated via ChatGPT/Codex OAuth without model not allowed from missing forward-compat/catalog synthesis.

Security Impact (required)

• New permissions/capabilities? No
• Secrets/tokens handling changed? No
• New/changed network calls? No
• Command/tool execution surface changed? No
• Data access scope changed? No

Repro + Verification

Environment

• Model/provider: openai-codex / gpt-5.4-pro (Codex OAuth, ChatGPT Pro plan as in issue).

Steps (from issue)

1. Authenticate with openai-codex OAuth (plan Pro).
2. Set or select model openai-codex/gpt-5.4-pro.
3. Before: model not allowed: openai-codex/gpt-5.4-pro.
4. After: model resolves via provider forward-compat and catalog augmentation.

Verification (local)

• pnpm test extensions/openai/openai-codex-provider.test.ts
• pnpm test src/agents/model-compat.test.ts -t "plugin-advertised"
• pnpm test src/auto-reply/thinking.test.ts -t listThinkingLevels
• pnpm test extensions/discord/src/monitor/native-command.think-autocomplete.test.ts
• pnpm test src/plugins/provider-runtime.test.ts -t "resolves bundled catalog hooks"
• pnpm test src/commands/models/list.list-command.forward-compat.test.ts -t "gpt-5.4-pro"

Evidence

• Unit tests updated/added for Codex gpt-5.4-pro resolution and aligned harnesses.

Human Verification (required)

• Verified: Static/code-path review; targeted Vitest runs above on the development branch.
• Not verified: Live ChatGPT Pro OAuth end-to-end call (no production tokens in this environment).

Compatibility / Migration

• Backward compatible? Yes (additive model id support).
• Config/env changes? No.
• Migration needed? No.

Failure Recovery (if this breaks)

• Revert the PR commit or pin default to openai-codex/gpt-5.4 in config until fixed.

Risks and Mitigations

• Risk: Upstream Codex API constraints (e.g. minimum thinking level) not enforced client-side.
  • Mitigation: Issue already notes optional follow-up; monitor user reports and add policy if needed.

Changed files

• extensions/discord/src/monitor/native-command.think-autocomplete.test.ts (modified, +6/-2)
• extensions/openai/openai-codex-provider.test.ts (modified, +93/-0)
• extensions/openai/openai-codex-provider.ts (modified, +41/-5)
• extensions/openai/shared.ts (modified, +10/-0)
• src/agents/model-compat.test.ts (modified, +3/-1)
• src/agents/pi-embedded-runner/model.provider-runtime.test-support.ts (modified, +17/-1)
• src/auto-reply/reply.directive.directive-behavior.e2e-harness.ts (modified, +2/-0)
• src/auto-reply/thinking.test.ts (modified, +2/-1)
• src/commands/models/list.list-command.forward-compat.test.ts (modified, +35/-0)
• src/plugins/provider-runtime.test-support.ts (modified, +1/-0)
• src/plugins/provider-runtime.test.ts (modified, +1/-0)