[Programcode]

This PR adds the Prepare tab, Generate, the generated case as a document (mock 03), Attack and Export, from plans/2026-10-08-native-ios-app.md. It targets feat/native-ios.

Prepare tab

• The tab lists the four tools. Generate works now; Brainstorm, Factoids and Flowchart show "on its way" until PRs 4b and 4c.
• Below the tools are the Current Debate's saved Prepare results, newest first.
• The Debate tab's tiles and its "Pick up where you left off" rows open the same screens.

Generate

• The form has the website's fields and defaults: In favor or Against, role, length, audience, tone and a style note. They are native pickers.
• If you type a different motion, the website's motion gate appears as a native alert: switch to that debate, create one, or use the motion once.
• Requests go through the helper Ask now shares with Generate and Attack, so they keep the same headers, the 200 s timeout and the abort before the session read.
• The model settings come from feature_configs for each tool. The debate context includes pins.
• The output runs through the website's own transform, which this PR moves into @debatelab/core. client.ts imports it back in one line, and its behaviour is unchanged.
• Saving matches the website:
  • The case is recorded like the website's recordResult.
  • Researched sources go to the debate's Auto-saved folder, and the "..." menu offers the website's "Undo auto-saved sources".
  • In one-off mode nothing is saved, and Save in the menu stores the case outside any debate.

The document

• The building blocks (framing card, numbered collapsible arguments, evidence quotes with citation chips, pinned actions) are written so 4b can reuse them.
• The native toolbar has back, share, and "..." with Discuss, Save, Undo auto-saved sources and Delete.
• Every field the website's GenerateResults shows also appears here. Fields the mock doesn't have go below the arguments in the same style: Case Structure, Predicted Opponent Case, Prepared Counters, Researched sources, and a footer with the model and duration.

Attack

• "Attack this case" opens Attack with the case text filled in, which is how the website hands a case over. Practice's Attack row also opens it now.
• Attacks are saved the way the website's AttackPage saves them.
• The pinned next step is the website's Discuss, which opens Ask with its message.

Export

• PDF goes through expo-print and the share sheet, using the website's PDF file names.
• Copy as text copies exactly what the website's Clipboard button copies.
• Word is left out. docx would have to run under Hermes in Expo Go, and that can't be tested on the VPS.
• Google Docs is left out. It needs Google sign-in, which comes with the dev build.

Checks

Run locally on Node 22, because GitHub Actions starts no jobs on this account (billing):

• npm run verify exits 0: 287 Vitest, 30 collaboration and 333 Deno tests, the build, the service-worker check and 30 Chromium smoke tests.
• Each of the five commits typechecks and lints on its own.
• The screenshot set has 224 screenshots signed in to staging: iPhone and iPad, light and dark, English and Hungarian. It includes the HTML handed to expo-print and the copied text.

Live on the staging seed Pro account:

• One Generate (67 s) and one Attack (86 s).
• The case opened from the Debate tab, exported to PDF HTML and copied as text.
• Both results were deleted through the app's menu, along with the 10 auto-saved sources the run created. 0 are left.

Not proven

These need Adam's phone:

• the native toolbar menus, action sheets and share sheet
• a real PDF file
• swipe-back

Discuss to Ask did not run live.

🤖 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 2:23pm UTC
debatelab-staging	Ready	Preview	Oct 9, 2026 2:23pm UTC

[Copilot]

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

[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: 8b1210a5-3714-4896-ad36-cb8ec958bf57

---

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]

Mock 03 against the app's web export (light English, dark Hungarian, iPad):

[Programcode]

Codex review (gpt-6.1-sol)

• r1 at 3d689ab: 7 findings. I verified both High ones myself.
  • High: form drafts and the open-results cache survived sign-out.
  • High: Undo auto-saved could delete sources belonging to another debate.
  • Medium: autosave kept running after Cancel.
  • Medium: Prepare showed the previous debate's results.
  • Medium: meta.truncated was dropped, so the shortening notice never showed.
  • Low: sourceAssessments was ignored.
  • Low: model-supplied ids were used as expansion keys.
  • All seven fixed in 7b939df:
    • Session memory is now cleared by the auth provider on every account change.
    • Undo checks that the source belongs to the current debate and filters by debate_id.
    • Autosave gets a cancel signal; results.test.ts fails without the fix.
    • Prepare is keyed by debate.
    • The website's shortening notice now shows.
    • Assessments are written to the debate captured with the request. The website's parser and merge moved to @debatelab/core, line for line.
    • Expansion keys are indexes.
• r2 at 7b939df: 2 Medium.
  • A late Retry on Prepare could overwrite newer state.
  • Assessment writes went through after Cancel.
  • Both fixed in adb53ae: loads now carry a sequence number, and the assessment write takes a signal. sources.test.ts fails without the fix.
• r3 at adb53ae: no findings. The r2 scenarios fail on the previous head and pass on this one.

Accepted: if a source moved to another debate, Undo's "Removed N" count can come out too high.

Local checks at adb53ae. GitHub Actions starts no jobs because of billing, so these ran locally:

• npm run verify exit 0, with 289 Vitest, 333 Deno and 30 Chromium smoke tests.
• Mobile typecheck, eslint and knip are clean.
• 224 screenshots pass at 7b939df. adb53ae doesn't change rendering.

Merging into feat/native-ios.

🤖 Generated with Claude Code