René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Thinking-level (reasoning effort) picker in the chat composer","articleBody":"## Problem or Motivation\n\nOpenWork supports six per-session **thinking levels** (`off` / `low` / `medium` /\n`high` / `xhigh` / `max`), but there is **no way to change the thinking level from\nthe chat composer**. Today the only place to influence it is Settings → AI, which\nsets a *per-workspace default* for new chats. To dial reasoning effort up for a\nsingle hard task or down for a quick one, you have to leave the conversation, open\nsettings, change the default, and start a new chat.\n\nComparable desktop apps put this control right in the composer, next to the model\npicker: Claude Code Desktop ships a dedicated **effort menu** (`⌘⇧E`) alongside its\nmodel menu, so you can raise or lower reasoning effort per message without leaving\nthe prompt.\n\nOpenWork already has the neighbor for it — the composer has a **model picker**\ndropdown — but the thinking-level equivalent is missing even though all the\nplumbing is already in place:\n\n- The composer (`FreeFormInput`) already receives `thinkingLevel` and\n  `onThinkingLevelChange` props — they are threaded from `ChatPage` /\n  `DraftChatPage` → `ChatDisplay` → `ChatInputZone` → `InputContainer` →\n  `FreeFormInput`, and both pages already persist the value (draft\n  `createOptions.thinkingLevel` / session options). But `FreeFormInput` never\n  renders a trigger for it — the prop is only read for an ACP command refresh.\n- The level metadata already exists: `THINKING_LEVELS` (id + `nameKey` +\n  `descriptionKey`) in `packages/shared/src/agent/thinking-levels.ts`, and the\n  `thinking.*` translation keys are already present in **all** locales.\n\nSo the value is fully wired end-to-end — the app just has no UI to set it from the\nplace you actually compose.\n\n## Proposed Solution\n\nAdd a **thinking-level picker** to the composer toolbar, immediately before the\nmodel picker, mirroring the existing model-picker dropdown:\n\n- A toolbar button showing the current level's name (e.g. \"Medium\") with a brain\n  icon and a chevron; tooltip \"Thinking\".\n- Opening it lists all six levels (`THINKING_LEVELS` order) with name +\n  description, a check on the active level, and calls the already-wired\n  `onThinkingLevelChange(level)` on select — so the choice persists exactly like a\n  model change does.\n- Hidden in `compactMode` (same rule the model picker uses).\n\nThis is a **frontend-only** change:\n\n- It renders a Radix dropdown (the same `StyledDropdownMenu*` primitives the model\n  picker uses) and calls the existing `onThinkingLevelChange` prop. No backend / no\n  qwen-code change — the value already flows to the runtime today.\n- It introduces **no new i18n keys**: level names/descriptions reuse the existing\n  `thinking.*` keys and the trigger tooltip reuses `settings.ai.thinking`.\n\n## Alternatives Considered\n\n- **Settings → AI thinking default** already exists, but it is a *workspace\n  default for new chats*, not a per-conversation control. You can't raise effort\n  for the current task without starting over.\n- **A keyboard-only shortcut** (à la `⌘⇧E`) could open a menu, but OpenWork's\n  established pattern for composer controls is a visible toolbar dropdown (the\n  model picker), so a visible picker is the more discoverable, consistent fit. A\n  shortcut can be layered on later.\n\n## Additional Context\n\nFeasibility: **frontend-only / doable.** Verified against `apps/electron` — the\n`thinkingLevel` / `onThinkingLevelChange` props already reach `FreeFormInput` and\nboth the draft and active chat pages already persist the value; only the UI trigger\nis missing.\n\n### Acceptance criteria (CDP e2e assertion)\n\nA new `e2e/assertions/thinking-level-picker.assert.ts` drives the real built app\nover CDP and asserts the full path:\n\n1. The composer renders a thinking-level trigger button.\n2. Clicking it opens a menu listing all **six** thinking levels.\n3. The trigger reflects a known current level (the default), and the active level\n   is marked as selected in the menu.\n4. Selecting a **different** level closes the menu **and** updates the trigger\n   label to the newly selected level — proving the picker actually changes the\n   session's thinking level, not merely displays a list.\n\nPart of the autonomous desktop-feature loop (`loop-bot`).","author":{"url":"https://github.com/DragonnZhang","@type":"Person","name":"DragonnZhang"},"datePublished":"2026-07-02T01:25:52.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":0},"url":"https://github.com/44/openwork/issues/44"}