feat(experiments): register PARALLEL_TOOL_EXECUTION flag (internal-only)

[edelauna]

Related GitHub Issue

Closes: #363

Description

Registers the PARALLEL_TOOL_EXECUTION experiment flag as an internal-only, preloaded flag (default false, showInSettings: false).

• Added "parallelToolExecution" to experimentIds and experimentsSchema in packages/types/src/experiment.ts
• Added PARALLEL_TOOL_EXECUTION: "parallelToolExecution" to EXPERIMENT_IDS and { enabled: false, showInSettings: false } to experimentConfigsMap in src/shared/experiments.ts
• Added showInSettings?: boolean to ExperimentConfig interface; ExperimentalSettings.tsx filters out configs where showInSettings === false, so the flag is wired up but invisible in the Settings UI until Epic 4 is ready
• Added unit tests asserting default-off behavior and that the flag does not render in the Settings panel

Test Procedure

nvm use 20.20.2
pnpm --filter zoo-code exec vitest run shared/__tests__/experiments.spec.ts
pnpm --filter @roo-code/vscode-webview exec vitest run src/components/settings/__tests__/ExperimentalSettings.spec.tsx

Both pass with no failures.

Pre-Submission Checklist

• Issue Linked: This PR is linked to an approved GitHub Issue (see "Related GitHub Issue" above).
• Scope: My changes are focused on the linked issue (one major feature/fix per PR).
• Self-Review: I have performed a thorough self-review of my code.
• Testing: New and/or updated tests have been added to cover my changes (if applicable).
• Documentation Impact: No documentation updates required — flag is internal-only and not visible in Settings.
• Contribution Guidelines: I have read and agree to the Contributor Guidelines.

Summary by CodeRabbit

• New Features

  • Introduced a new experimental flag (parallelToolExecution) and updated the experiment definitions and validation.
  • Added settings visibility control so certain experiments can be hidden from the UI by default.

• Bug Fixes

  • Improved experimental settings rendering to show only flags meant to appear in settings.

• Tests

  • Extended experiment-related unit tests to cover the new flag, default behavior, and settings visibility.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 20bfc856-e821-41cc-b0c0-056109242b79

📥 Commits

Reviewing files that changed from the base of the PR and between eb2089a and 4367bf3.

📒 Files selected for processing (2)

• src/shared/experiments.ts
• webview-ui/src/components/settings/__tests__/ExperimentalSettings.spec.tsx

🚧 Files skipped from review as they are similar to previous changes (2)

• webview-ui/src/components/settings/tests/ExperimentalSettings.spec.tsx
• src/shared/experiments.ts

---

📝 Walkthrough

Walkthrough

Adds a new parallelToolExecution experiment flag to the Zod type schema, registers it in experimentConfigsMap with enabled: false and showInSettings: false, introduces a showInSettings optional field on ExperimentConfig, filters out configs with showInSettings === false from the settings UI, and adds corresponding tests.

Changes

PARALLEL_TOOL_EXECUTION Experiment Flag

Layer / File(s)	Summary
Zod schema and experimentIds type contract packages/types/src/experiment.ts	experimentIds array and experimentsSchema are extended with parallelToolExecution, updating the inferred ExperimentId and Experiments types.
Experiment config entry, showInSettings field, and UI filter src/shared/experiments.ts, webview-ui/src/components/settings/ExperimentalSettings.tsx	PARALLEL_TOOL_EXECUTION is added to EXPERIMENT_IDS and experimentConfigsMap with enabled: false and showInSettings: false. ExperimentConfig gains an optional showInSettings field. The settings component filter is extended to exclude entries where showInSettings === false.
Experiment and settings component tests src/shared/__tests__/experiments.spec.ts, webview-ui/src/components/settings/__tests__/ExperimentalSettings.spec.tsx	Existing mock objects are updated to include parallelToolExecution: false. A new describe("PARALLEL_TOOL_EXECUTION") block validates ID wiring, config shape, and isEnabled behavior. A new ExperimentalSettings component test asserts that internal-only flags are absent from the rendered DOM.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Poem

🐰 A new flag hops into the map,
parallelToolExecution fills the gap.
Hidden from settings, a quiet internal guest,
showInSettings: false keeps it under arrest.
The schema expands, the tests all agree —
this rabbit's experiment runs silently! 🌿

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely summarizes the main change: registering the PARALLEL_TOOL_EXECUTION flag as an internal-only feature.
Description check	✅ Passed	The description fully follows the template with linked issue, clear explanation of implementation details, test procedures with commands, and completed pre-submission checklist.
Linked Issues check	✅ Passed	The PR meets all acceptance criteria from #363: defaults to false, extends experimentsSchema and experimentConfigsMap, adds showInSettings property, includes tests verifying default-off behavior, and filters Settings UI appropriately.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to #363 requirements: registering the flag infrastructure, adding type safety, filtering Settings UI, and adding comprehensive unit tests. No extraneous modifications detected.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch issue/363

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[navedmerchant]

Looks good! couple of comments below

[navedmerchant]

[Low] Document the showInSettings contract

The showInSettings !== false filter in ExperimentalSettings.tsx relies on undefined !== false being true, so all pre-existing configs (which omit the field) remain visible. This is correct and backward-compatible, but the implicit "undefined = visible" contract isn't documented here.

Impact: Maintainability — a future contributor could misread the optional field and assume omission hides the flag.

Suggestion: Add a one-line JSDoc, e.g. /** Defaults to true; set to false to hide from the Settings panel. */, so the convention is self-documenting.

[navedmerchant]

[Low] Translation key TODO for when visibility is enabled

No translation key (settings:experimental.PARALLEL_TOOL_EXECUTION.name / .description) was added to en/settings.json. This is harmless now because the flag is filtered out before render, so the key is never looked up.

Impact: None today. When "Epic 4" flips showInSettings to true (or removes the filter), the UI will render the raw key string instead of a human-readable label until translations are added. The repo's scripts/find-missing-i18n-key.js scans for referenced keys, and since the key is never referenced while hidden, it won't catch this gap at that transition.

Suggestion: Leave a TODO anchored to this flag (or note in issue #363) that translation entries must be added in the same PR that enables showInSettings: true.

[navedmerchant]

[Low] Optional: assert special-cased entries still render

The test asserts PREVENT_FOCUS_DISRUPTION and RUN_SLASH_COMMAND render and PARALLEL_TOOL_EXECUTION does not, but doesn't assert that the special-cased IMAGE_GENERATION and CUSTOM_TOOLS entries still render. Those branches are exercised by the same .filter pipeline, so a regression that accidentally hid them wouldn't be caught here.

Impact: Low — test coverage gap. The test's stated intent ("does not render internal-only experiment flags") implicitly promises that non-internal flags do render, but only the generic-path flags are checked.

Suggestion: Optionally add assertions that the IMAGE_GENERATION and CUSTOM_TOOLS labels (or their rendered wrappers) are present, so the test guards the full filter behavior rather than a subset.