CLI: include synthetic Codex gpt-5.4 rows in models --all by vicjayjay · Pull Request #38696 · openclaw/openclaw

vicjayjay · 2026-03-07T08:39:05Z

Summary

Describe the problem and fix in 2-5 bullets:

Problem: openclaw models list --all only used discovered registry models, so synthetic forward-compat catalog entries like openai-codex/gpt-5.4 were omitted even though source-level fallback support already existed.
Why it matters: users comparing docs/config/auth output against models list --all saw inconsistent Codex OAuth model coverage.
What changed: merge synthetic catalog entries into models list --all via loadModelCatalog() and resolveModelWithRegistry(), and update/add tests for the openai-codex/gpt-5.4 path.
What did NOT change (scope boundary): this PR does not change provider auth flows or model runtime behavior beyond list output and stale test expectations.

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

User-visible / Behavior Changes

openclaw models list --all now includes synthetic forward-compat entries like openai-codex/gpt-5.4 when the catalog exposes them.
openclaw models auth login --provider openai-codex test expectations now match the current gpt-5.4 default.

Security Impact (required)

New permissions/capabilities? (Yes/No) No
Secrets/tokens handling changed? (Yes/No) No
New/changed network calls? (Yes/No) No
Command/tool execution surface changed? (Yes/No) No
Data access scope changed? (Yes/No) No
If any Yes, explain risk + mitigation:

Repro + Verification

Environment

OS: Windows 11
Runtime/container: Node 22 / pnpm / Vitest
Model/provider: openai-codex/gpt-5.4
Integration/channel (if any): None
Relevant config (redacted): default OpenClaw config with Codex OAuth model refs in tests

Steps

Configure or synthesize openai-codex/gpt-5.4 via the source-level model catalog fallback.
Run openclaw models list --all.
Run the targeted auth + model list tests.

Expected

models list --all includes the synthetic openai-codex/gpt-5.4 row.
Auth login tests reflect the current gpt-5.4 default.

Actual

Before this patch, models list --all could omit openai-codex/gpt-5.4, and auth tests still expected gpt-5.3-codex.

Evidence

Attach at least one:

Failing test/log before + passing after
Trace/log snippets
Screenshot/recording
Perf numbers (if relevant)

Human Verification (required)

What you personally verified (not just CI), and how:

Verified scenarios: targeted Vitest run for src/commands/models/auth.test.ts and src/commands/models/list.list-command.forward-compat.test.ts.
Edge cases checked: configured-mode fallback still marks synthetic Codex rows as available when auth exists; configured-mode error handling still exits cleanly when the registry is unavailable.
What you did not verify: full end-to-end OAuth login in an interactive TTY session.

Compatibility / Migration

Backward compatible? (Yes/No) Yes
Config/env changes? (Yes/No) No
Migration needed? (Yes/No) No
If yes, exact upgrade steps:

Failure Recovery (if this breaks)

How to disable/revert this change quickly: revert commit b2d3ebf0a.
Files/config to restore: src/commands/models/list.list-command.ts, src/commands/models/auth.test.ts, src/commands/models/list.list-command.forward-compat.test.ts.
Known bad symptoms reviewers should watch for: duplicate rows or missing rows in models list --all for forward-compat catalog entries.

Risks and Mitigations

List only real risks for this PR. Add/remove entries as needed. If none, write None.

Risk: synthetic catalog entries could surface duplicate rows if merge keys are inconsistent.
- Mitigation: rows are deduplicated by modelKey(provider, id) before printing, and the new test covers the synthetic openai-codex/gpt-5.4 path.

greptile-apps · 2026-03-07T08:43:13Z

Greptile Summary

This PR merges synthetic forward-compat catalog entries (e.g., openai-codex/gpt-5.4) into models list --all output by calling loadModelCatalog() after the registry load and resolving each catalog entry via resolveModelWithRegistry(), and updates auth test expectations to reflect the current gpt-5.4 default.

Key changes:

list.list-command.ts: Adds a catalog-merge block (gated on opts.all && modelRegistry) that deduplicates by modelKey and resolves synthetic entries into full Model objects before they reach the sort/print loop.
list.list-command.forward-compat.test.ts: Adds a mock for loadModelCatalog and a new test verifying the synthetic row is present in --all output; also fixes the default config entry from gpt-5.3-codex to gpt-5.4.
auth.test.ts: Updates two assertions from gpt-5.3-codex to gpt-5.4 to match current behavior.

Issue found: The toModelRow calls inside the --all loop do not pass allowProviderAvailabilityFallback: true for catalog-merged entries. Because synthetic models won't appear in availableKeys (populated from registry.getAvailable()), they will always render as available: false in --all output even when the user has auth configured for the provider. The configured-mode path already handles this correctly using !discoveredKeys.has(key). The new test does not assert on the available field, leaving this gap undetected.

Confidence Score: 3/5

Safe to merge as a display-only change, but synthetic entries in --all will consistently show as unavailable even when the user is authenticated, which is misleading and inconsistent with the configured-mode behavior.
The core mechanic (merging catalog entries into the model list) is correct and the deduplication logic is sound. The resolveForwardCompatModel path always returns a model with the requested id, so no key mismatch occurs. However, omitting allowProviderAvailabilityFallback for catalog-merged --all rows means these rows will show available: false even for authenticated users — the opposite of the behavior described in the PR for configured mode. This is a functional inaccuracy in the displayed output and warrants a fix before merge.
Pay close attention to src/commands/models/list.list-command.ts lines 98–118 and the corresponding test in list.list-command.forward-compat.test.ts.

Comments Outside Diff (1)

src/commands/models/list.list-command.ts, line 98-118 (link)

Catalog-merged rows always show available: false in --all mode

Synthetic catalog entries resolved via resolveModelWithRegistry (forward-compat fallbacks like openai-codex/gpt-5.4) will never appear in availableKeys — those keys come from registry.getAvailable(), which only knows about models the registry itself discovered. As a result, the toModelRow call here never sets allowProviderAvailabilityFallback: true, so when availableKeys is defined the available field is forced to false for every catalog-merged row, even when the user is authenticated for the provider.

The configured-mode path (the else branch, lines 130–155) already handles this correctly by passing allowProviderAvailabilityFallback: !discoveredKeys.has(modelKey(model.provider, model.id)). Because discoveredKeys is computed before the catalog merge, it still accurately reflects only the originally-registered models, making it the right key to check here as well.

Adding allowProviderAvailabilityFallback: !discoveredKeys.has(key) to the toModelRow call inside this loop would bring --all mode in line with configured mode for synthetic/forward-compat entries.

The new test only asserts on key presence and does not exercise the available field, so this regression is not caught. A companion test that mocks a provider auth profile and asserts available: true on the synthetic --all row would cover the gap.

_{Last reviewed commit: b2d3ebf}

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: b2d3ebf0a8

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

chatgpt-codex-connector · 2026-03-07T08:43:58Z

src/commands/models/list.list-command.ts

+      mergedByKey.set(key, model);
+    }
+
+    models = [...mergedByKey.values()];


Preserve auth-based availability for synthetic --all rows

Merging catalog-only entries into models here introduces rows that are not present in availableKeys, but the later --all rendering path still calls toModelRow without allowProviderAvailabilityFallback; per toModelRow’s logic this forces available to false whenever getAvailable() lacks that key. In practice, a synthetic forward-compat model like openai-codex/gpt-5.4 can be usable (auth exists and non---all mode marks it available) yet models list --all will incorrectly report it as unavailable.

Useful? React with 👍 / 👎.

dorukardahan · 2026-03-08T18:36:39Z

i opened #40160 as a smaller follow-up on latest main. it keeps discovered --all rows, fills the missing synthetic catalog rows, and covers the availability + empty-catalog cases. i think it supersedes this one: #40160

Upstream refactored --all listing into appendDiscoveredRows/appendCatalogSupplementRows and expanded forward-compat tests. Accept upstream's structure since the feature is already covered by the new test suite. Co-Authored-By: Claude Opus 4.6 (1M context) <[email protected]>

steipete · 2026-04-04T01:47:46Z

Thanks for the patch. I pulled latest main and checked the current tree; this is already covered.

Current main already has explicit coverage for this path in src/commands/models/list.list-command.forward-compat.test.ts:

:219-235 marks synthetic openai-codex/gpt-5.4 rows as available when provider auth exists
:263-290 covers models list --all including synthetic Codex gpt-5.4 rows when the catalog supports them

Closing as already in main. Thanks for chasing this.

CLI: list Codex gpt-5.4 in models --all

b2d3ebf

openclaw-barnacle bot added commands Command implementations size: S labels Mar 7, 2026

chatgpt-codex-connector bot reviewed Mar 7, 2026

View reviewed changes

dorukardahan mentioned this pull request Mar 8, 2026

fix(models): keep --all aligned with synthetic catalog rows #40160

Merged

openclaw-barnacle bot added size: XS and removed size: S labels Mar 18, 2026

steipete closed this Apr 4, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

CLI: include synthetic Codex gpt-5.4 rows in models --all#38696

CLI: include synthetic Codex gpt-5.4 rows in models --all#38696
vicjayjay wants to merge 2 commits intoopenclaw:mainfrom
vicjayjay:main

vicjayjay commented Mar 7, 2026

Uh oh!

greptile-apps bot commented Mar 7, 2026 •

edited

Loading

Comments Outside Diff (1)

Uh oh!

chatgpt-codex-connector bot left a comment

Uh oh!

chatgpt-codex-connector bot Mar 7, 2026

Uh oh!

dorukardahan commented Mar 8, 2026

Uh oh!

steipete commented Apr 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

vicjayjay commented Mar 7, 2026

Summary

Change Type (select all)

Scope (select all touched areas)

Linked Issue/PR

User-visible / Behavior Changes

Security Impact (required)

Repro + Verification

Environment

Steps

Expected

Actual

Evidence

Human Verification (required)

Compatibility / Migration

Failure Recovery (if this breaks)

Risks and Mitigations

Uh oh!

greptile-apps bot commented Mar 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Greptile Summary

Confidence Score: 3/5

Comments Outside Diff (1)

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

chatgpt-codex-connector bot Mar 7, 2026

Choose a reason for hiding this comment

Uh oh!

dorukardahan commented Mar 8, 2026

Uh oh!

steipete commented Apr 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

greptile-apps bot commented Mar 7, 2026 •

edited

Loading