[Programcode]

Last piece of milestone 1 from plans/2026-10-08-native-ios-app.md: the Library tab to mock 04 and Ask to mock 07, with streaming answers. Stacked on #125. GitHub retargets this PR to feat/native-ios when #125 merges.

What the app now does:

• Library. It lists the Current Debate's sources, with search and the All, Files, Links and Notes filters. The plus menu can paste a link, which goes through the import-source Edge Function, or write a note. Both are saved with the website's columns and folder. Tapping a row opens the source read-only, and its page opens in the in-app browser. Swiping a row left deletes it after a confirmation. "Choose a file" and "Scan pages" are left out of this PR.
• Practice and Ask. Practice lists the website's practice tools. Ask opens full screen; the other tools say they are on their way.
  • Ask sends the same request as the website. A plain question streams from ai-stream through expo/fetch. With Citations, Web search, Pros/Cons or Argument map turned on, the full answer comes from ai, because ai-stream doesn't accept those options.
  • The model, settings and Current Debate context, pins included, are chosen the way the website chooses them. The Pro check and default model settings come from @debatelab/core, the same code Add the Debate tab, debate switcher and You tab to the app #125 uses.
  • Finished exchanges are saved to chats and messages. Past chats reopen from the history sheet. Leaving mid-answer cancels the request and saves nothing partial.
  • Answers render as Markdown, with citation chips that open the Library source or the cited page.
• Shared code. The SSE parser moved from src/lib/ai/streaming.ts to packages/core/src/ai/sse.ts. The website imports it back, so its behaviour and tests are unchanged. That is the only website file this PR touches.
• Ownership. Every read of a user-owned table also filters on user_id, as Add the Debate tab, debate switcher and You tab to the app #125 does. RLS still enforces ownership.

Checks, run locally on Node 22 at 685f4fb, because GitHub Actions on this account starts no jobs (billing):

• At a6830d0, the same code rebased onto Add the Debate tab, debate switcher and You tab to the app #125's previous head: npm run verify exit 0 (286 Vitest tests, collaboration, Deno edge tests, build, service worker, 30 Chromium smoke), plus docs:check, the web export and 152 screenshots of the web export signed in to staging. At 685f4fb: typecheck, lint and all 286 tests pass.

• Live on staging with throwaway 3c-test- data:

  • A link and a note added, then deleted.
  • Three real Ask requests. A plain question streamed from ai-stream, with the first token at 6.5 s. A Citations request came back from ai with 6 citation buttons. A request stopped mid-stream saved nothing partial.
  • The chat was saved and reopened from history.

  All test sources, chats and messages were deleted afterwards (0 left).

• Every native module is in the App Store build of Expo Go.

Not proven: native rendering on an iPhone (sheets, swipe-to-delete, keyboard, the in-app browser). Those are for the milestone 1 check on Adam's phone.

🤖 Generated with Claude Code

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
debate-navigator	Ready	Preview	Oct 9, 2026 11:59am UTC
debatelab-staging	Ready	Preview	Oct 9, 2026 11:59am UTC

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[Programcode]

Mocks against the app's web export, then real Ask on staging (a cited answer, and the saved chat reopened from history):

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

• Configuration used: Repository YAML (base), Organization UI (inherited)
• Review profile: CHILL
• Plan: Free
• Run ID: ac4abf41-a529-4a4d-b835-29e16e02b268

---

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

---

Note

🎁 Summarized by CodeRabbit Free

Your organization is on the Free plan. CodeRabbit will generate a high-level summary and a walkthrough for each pull request. For a comprehensive line-by-line review, please upgrade your subscription to CodeRabbit Essentials by visiting https://app.coderabbit.ai/settings/billing.

---

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

[Programcode]

Codex review (gpt-6.1-sol)

• r1 at 685f4fb
  • Medium. Leaving Ask while the session token was refreshing still sent the request, which used up quota. Fixed in 41a0b68: the abort is wired up before the session read and checked again after it. apps/mobile/src/lib/ask.test.ts fails on the old code.
  • Medium. If the messages insert failed, the new chat id was dropped, so the next save created a duplicate chat. Fixed in 41a0b68: ChatSaveError carries the record, and Ask keeps it.
  • Low. The app showed the raw backend import error in translated UI. Fixed in 41a0b68: it now shows only translated text and logs the raw message. The website's LibraryPage.tsx does the same thing and is tracked separately; this PR leaves it alone.
  • r1 also found SSE extraction matches the original across 16,997 chunk-boundary cases.
• r2 at 41a0b68: no findings. A control probe confirmed the old abort bug.

Local checks at 41a0b68. GitHub Actions starts no jobs on this account because of billing, so these ran locally:

• npm run verify exit 0, including 30 Chromium smoke tests.
• Lint, typecheck, npm test (287), docs:check, build, the web export, and 152 screenshots against staging.

Merging into feat/native-ios.

🤖 Generated with Claude Code