Summary

• remove the remaining inline fix wrapper from lib/codex-manager.ts
• route both CLI dispatch and the login-menu auto-fix action directly through the extracted fix command module

What Changed

• deleted the inline runFix() wrapper
• updated codex auth fix dispatch to call runFixCommand(...) directly
• updated the login-menu fix action to use the same extracted command path

Validation

• npm run test -- test/codex-manager-fix-command.test.ts test/codex-manager-cli.test.ts
• npm run lint
• npm run typecheck
• npm run build

Risk and Rollback

• Risk level: low
• Rollback plan: revert 987fca2 to restore the inline fix wrapper

Additional Notes

• this completes the direct-routing cleanup for the extracted manager command modules in the current train

note: greptile review for oc-chatgpt-multi-auth. cite files like lib/foo.ts:123. confirm regression tests + windows concurrency/token redaction coverage.

Greptile Summary

this pr completes the direct-routing cleanup for the fix command by replacing the deleted runFix() wrapper with a buildFixCommandDeps() factory — directly addressing the duplicate-deps hazard flagged in the previous review. both the login-menu auto-fix path and the CLI dispatch now call runFixCommand through the shared factory, keeping deps DRY. the runDoctor() wrapper is also removed and its deps are inlined at the single dispatch site.

key changes:

• runFix() wrapper deleted; replaced with buildFixCommandDeps() factory used at both call sites
• runDoctor() wrapper deleted; doctor deps inlined directly at command === "doctor" dispatch (single call site, so no factory added)
• type FixCommandDeps imported to satisfy the factory's return type annotation
• no logic changes — this is a pure structural refactor

issues found:

• buildFixCommandDeps() has no dedicated vitest coverage; the fix-command test uses its own mock factory and the cli test mocks the full module, so the production wiring is exercised only in e2e — worth a smoke test given the 27-dep surface area and the 80% threshold requirement
• doctor deps are inlined without a factory, creating an asymmetry with fix; not a bug today (one call site), but a future maintenance hazard if a doctor menu action is ever added

Confidence Score: 4/5

• safe to merge — pure structural refactor with no logic changes, correct factory wiring, and existing integration coverage on the fix dispatch path
• the factory correctly captures all 27 deps from the deleted runFix() wrapper (typescript enforces completeness at compile time), both call sites are updated, and the cli integration test exercises the fix dispatch path. score is 4 rather than 5 because buildFixCommandDeps() is not directly unit-tested and the doctor inline creates a pattern asymmetry that could become a real bug if doctor gains a second call site
• lib/codex-manager.ts — specifically the buildFixCommandDeps factory (no unit test) and the inlined doctor deps (no factory)

Important Files Changed

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    CLI["runCodexMultiAuthCli(rawArgs)"]
    MENU["runAuthLogin → login-menu loop"]

    CLI -->|"command === 'fix'"| FIX_DISPATCH["buildFixCommandDeps()"]
    MENU -->|"menuResult.mode === 'fix'"| FIX_MENU["buildFixCommandDeps()"]

    FIX_DISPATCH --> RFC["runFixCommand(rest, deps)"]
    FIX_MENU --> RFC2["runFixCommand(['--live'], deps)"]

    CLI -->|"command === 'doctor'"| DOC_INLINE["inline DoctorCommandDeps object"]
    DOC_INLINE --> RDC["runDoctorCommand(rest, deps)"]

    style FIX_DISPATCH fill:#22c55e,color:#fff
    style FIX_MENU fill:#22c55e,color:#fff
    style DOC_INLINE fill:#f59e0b,color:#fff

This is a comment left during a code review.
Path: lib/codex-manager.ts
Line: 2807-2840

Comment:
**no dedicated unit test for `buildFixCommandDeps`**

`buildFixCommandDeps()` wires 27 production functions and `DEFAULT_DASHBOARD_DISPLAY_SETTINGS` — but neither `test/codex-manager-fix-command.test.ts` nor `test/codex-manager-cli.test.ts` instantiates this factory directly. the fix-command test uses its own `createDeps()` mock, so a mis-wired dep (e.g. wrong display settings object, swapped function ref) would be invisible to vitest until an e2e run.

the cli test at line 1513 does exercise the `fix` dispatch path, but it mocks the entire module, so `buildFixCommandDeps()` itself never executes in the test suite.

suggest adding a smoke test that calls `buildFixCommandDeps()` and asserts structural completeness (e.g. all keys of `FixCommandDeps` are present and are functions / correct type), keeping the 80% threshold meaningful for this new code path.

How can I resolve this? If you propose a fix, please make it concise.

---

This is a comment left during a code review.
Path: lib/codex-manager.ts
Line: 3574-3601

Comment:
**doctor deps inlined — asymmetric with fix**

fix got `buildFixCommandDeps()` to prevent future copy-paste divergence; doctor's 20-dep object is now inlined at the single dispatch site with no factory. that's fine today (one call site), but if a doctor auto-fix action is ever added to the login-menu (the same pattern that motivated the fix factory), you'll be back to the same divergence risk.

keeping the pattern consistent now is low-effort:

```typescript
function buildDoctorCommandDeps(): DoctorCommandDeps {
    return {
        setStoragePath,
        getStoragePath,
        // ... all 20 deps
        setCodexCliActiveSelection,
    };
}
```

not a blocker, but worth tracking before the train closes.

How can I resolve this? If you propose a fix, please make it concise.

_{Last reviewed commit: "refactor: share fix ..."}