🤖 tests: stabilize Storybook fixtures for Pixel flake reduction

[ammar-agent]

Summary

Harden Storybook fixtures that used live Date.now() / Math.random() so Pixel baselines are less flaky across PR builds. Opened separately from #3807 after investigating its pending Pixel review.

Background

#3807 showed Pixel / Review: 5 snapshots awaiting review even though its product UI change was limited to OpenAI store-toggle save behavior, and ProvidersSection stories are explicitly PIXEL_DISABLED.

Findings:

• Many open PRs currently have pending Pixel reviews (2–109 snapshots), so this is not unique to 🤖 feat: default Grok Responses to store=false for ZDR parity #3807.
• Main’s latest Pixel build is clean (368 snapshots approved, auto-reviewed).
• 🤖 feat: default Grok Responses to store=false for ZDR parity #3807 itself is already merged to main; the lingering pending status is on the pre-merge PR head build.
• Pixel UI requires GitHub org sign-in, so exact story names for build 🤖 fix: improve web_fetch HTTP error reporting with body parsing #747 could not be scraped without interactive auth.

Likely flake contributors found in story fixtures: relative timestamps / random module tokens that can still affect provider tree keys or labels when capture paths bypass the global Date.now stub or when modules are re-evaluated.

Implementation

Replace nondeterministic time/random sources in Storybook stories with the shared NOW / STABLE_TIMESTAMP fixtures (src/browser/stories/storyTime.ts):

• LeftSidebar module render token
• BackgroundProcesses start times
• ProjectSidebar createdAt fixtures
• MemoryTab / MemorySection access times
• GoalTab / RightSidebar timestamps
• TranscriptDensity active tool start
• BashToolCall executing timer start
• MCP modal/settings testedAt fixtures

No production UI code changed.

Validation

• make typecheck
• make static-check
• Diff is stories-only; Pixel on this PR should either be clean or present a much smaller, reviewable set of real diffs.

Risks

• Low. Snapshot baselines may update once if previous “flaky” timestamps were accidentally baselined; that is intended.
• Does not auto-approve existing pending Pixel reviews on other PRs; those still need human approve/reject or a rebase onto a clean main baseline.

---

Generated with mux • Model: xai:grok-4.5 • Thinking: high

[ammar-agent]

@codex review

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 9ba144b4ea

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".

[ammar-agent]

@codex review

Addressed the three P2s:

1. Bash executing timer uses NOW - 1_000 (keeps ~1s state).
2. TranscriptDensity active bundle uses NOW - 39_000 (keeps 39s, not 1m39s).
3. LeftSidebar module token stays stable in Pixel/CI, but still bumps via import.meta.hot.data during Vite HMR so provider trees remount after story edits.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep them coming!

Reviewed commit: 1cf46354a8

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".