[Programcode]

Summary

• route every real AI request through OpenRouter
• make z-ai/glm-5.3-flash the only selectable/default model
• migrate legacy settings and remove stored client provider credentials
• use OpenRouter reported usage cost and keep cancelled work non-billable
• update pricing and privacy copy in English, Spanish, and Hungarian

Verification

• npm run verify on Node 22: 1,018 Vitest, 172 Edge, build, 134 Chromium
• pgTAP: 27 passed
• WebKit: 67 passed
• release reviewer: no remaining Critical or Important findings

Summary by CodeRabbit

• New Features

  • AI requests now use OpenRouter with GLM 5.3 Flash as the available model.
  • Pro users can choose low, medium, or high reasoning effort; trial users remain on low reasoning.
  • Model and reasoning settings are validated and migrated to supported defaults.
  • Usage and provider-reported costs are tracked, with incomplete costs shown as unknown.

• Documentation

  • Updated pricing, privacy, FAQ, upgrade, and setup content across supported languages to reflect the new AI service and limits.

[vercel]

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

Project	Deployment	Actions	Updated
debate-navigator	Ready	Preview	Sep 18, 2026 3:56pm UTC

[coderabbitai]

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Free

Run ID: aab73fdf-1cc1-4d85-bb67-88ea265fd057

📥 Commits

Reviewing files that changed from the base of the PR and between da14bc9 and d87759f.

📒 Files selected for processing (37)

• docs/architecture.md
• package.json
• src/components/ModelSelector.tsx
• src/lib/ai/client.ts
• src/lib/ai/providers/types.ts
• src/lib/ai/streaming.ts
• src/lib/stores/settings-store.ts
• src/locales/en.json
• src/locales/es.json
• src/locales/hu.json
• src/pages/AskPage.tsx
• src/test/ai/client-analytics.test.ts
• src/test/ai/client-branches.test.ts
• src/test/ai/client.test.ts
• src/test/ai/sparring.test.ts
• src/test/ai/streaming-integration.test.ts
• src/test/ai/streaming.test.ts
• src/test/components/ModelSelector.test.tsx
• src/test/observability.test.ts
• src/test/providers.test.ts
• src/test/security-hardening.test.ts
• src/test/settings-store.test.ts
• src/test/stores/settings-store-additional.test.ts
• src/test/stores/settings-store-branches.test.ts
• src/test/stores/settings-store-extended.test.ts
• supabase/README.md
• supabase/functions/_shared/cost.test.ts
• supabase/functions/_shared/cost.ts
• supabase/functions/_shared/model-catalog.test.ts
• supabase/functions/_shared/model-catalog.ts
• supabase/functions/_shared/research.test.ts
• supabase/functions/_shared/research.ts
• supabase/functions/ai-stream/handler_test.ts
• supabase/functions/ai-stream/index.ts
• supabase/functions/ai/generate-honesty_test.ts
• supabase/functions/ai/handler-research.test.ts
• supabase/functions/ai/index.ts

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The change routes AI requests through OpenRouter with z-ai/glm-5.3-flash, adds catalog-based reasoning controls, normalizes settings and migrations, records provider-reported costs, and updates streaming behavior, documentation, localization, and tests.

Changes

OpenRouter AI integration

Layer / File(s)	Summary
Shared model and settings contracts src/lib/ai/providers/types.ts, src/lib/stores/settings-store.ts, supabase/functions/_shared/model-catalog.ts, src/components/ModelSelector.tsx	The shared catalog defines the GLM model, token limits, and reasoning efforts. Settings normalize legacy values. The selector exposes reasoning controls for Pro subscriptions.
AI proxy and streaming supabase/functions/ai/index.ts, supabase/functions/ai-stream/index.ts, src/lib/ai/client.ts, src/lib/ai/streaming.ts	AI and streaming requests use OpenRouter, validate catalog values, forward resolved reasoning effort, and handle OpenRouter responses and stream cleanup.
Usage accounting and configuration supabase/functions/_shared/cost.ts, package.json, supabase/README.md, docs/architecture.md, src/locales/*.json	Usage normalization records valid provider costs and returns unknown cost for incomplete usage. Runtime configuration and product text now identify OpenRouter and the GLM model.
Validation coverage supabase/functions/**/*test.ts, src/test/**/*.test.*	Tests cover routing, model validation, reasoning tiers, settings migration, streaming failures and cancellation, telemetry, and reported-cost handling.

Estimated code review effort: 5 (Critical) | ~90 minutes

Sequence Diagram(s)

sequenceDiagram
  participant Client
  participant AIProxy
  participant ModelCatalog
  participant OpenRouter
  Client->>AIProxy: Submit model and reasoningEffort
  AIProxy->>ModelCatalog: Validate model and resolve effort
  ModelCatalog-->>AIProxy: Return limits and effective effort
  AIProxy->>OpenRouter: Send chat completion request
  OpenRouter-->>AIProxy: Return response and usage
  AIProxy-->>Client: Return AI result

Loading

---

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.}

[Copilot]

🟡 Changes recommended

Add the missing reasoning translations, OpenRouter provider-error classification, and streaming usage option.

Get a fresh assessment by requesting another Copilot review.

Pull request overview

Routes AI inference through OpenRouter with GLM 5.3 Flash, reasoning controls, reported-cost accounting, settings migration, and localized copy updates.

Changes:

• Centralizes OpenRouter model routing and reasoning configuration.
• Adds cancellation-aware usage and cost handling.
• Updates settings, tests, documentation, and translations.

File summaries

File	Summary
supabase/README.md	Documents OpenRouter configuration.
supabase/functions/ai/index.ts	Routes non-streaming inference through OpenRouter.
supabase/functions/ai/handler-research.test.ts	Tests routing, usage, and cancellation.
supabase/functions/ai/generate-honesty_test.ts	Updates provider test fixtures.
supabase/functions/ai-stream/index.ts	Implements streaming and cost accounting; usage option remains needed.
supabase/functions/ai-stream/handler_test.ts	Tests streaming behavior.
supabase/functions/_shared/research.ts	Propagates reasoning controls.
supabase/functions/_shared/research.test.ts	Updates research fixtures.
supabase/functions/_shared/model-catalog.ts	Defines model and reasoning policies.
supabase/functions/_shared/model-catalog.test.ts	Tests catalog behavior.
supabase/functions/_shared/cost.ts	Handles reported costs.
supabase/functions/_shared/cost.test.ts	Tests cost accounting.
src/test/stores/settings-store-extended.test.ts	Tests settings defaults.
src/test/stores/settings-store-branches.test.ts	Tests settings migrations.
src/test/stores/settings-store-additional.test.ts	Tests settings branches.
src/test/settings-store.test.ts	Tests canonicalization and persistence.
src/test/security-hardening.test.ts	Tests model allowlisting.
src/test/providers.test.ts	Tests the provider catalog.
src/test/observability.test.ts	Updates telemetry fixtures.
src/test/components/ModelSelector.test.tsx	Tests model and reasoning selection.
src/test/ai/streaming.test.ts	Updates streaming helper tests.
src/test/ai/streaming-integration.test.ts	Tests reasoning payload forwarding.
src/test/ai/sparring.test.ts	Updates sparring fixtures.
src/test/ai/client.test.ts	Tests client routing and reasoning.
src/test/ai/client-branches.test.ts	Updates client branch coverage.
src/test/ai/client-analytics.test.ts	Updates provider analytics expectations.
src/pages/AskPage.tsx	Uses OpenRouter result metadata.
src/locales/hu.json	Updates Hungarian copy.
src/locales/es.json	Updates Spanish copy.
src/locales/en.json	Updates English copy; reasoning labels need keys.
src/lib/stores/settings-store.ts	Migrates legacy settings and removes credentials.
src/lib/ai/streaming.ts	Updates streaming support and payloads.
src/lib/ai/providers/types.ts	Defines the OpenRouter provider catalog.
src/lib/ai/client.ts	Sends OpenRouter configuration; classifier needs OpenRouter support.
src/components/ModelSelector.tsx	Adds model and reasoning selectors.
package.json	Updates Edge test permissions.
docs/architecture.md	Documents routing and billing behavior.

Review details

Suppressed comments (3)

src/components/ModelSelector.tsx:23

• The catalog declares GLM 5.3 Flash supports only low and high reasoning (supabase/functions/_shared/model-catalog.ts:19), but this list exposes medium to Pro users. Selecting it persists and sends medium, which the server silently coerces to low, so the UI displays a setting that is not honored. Remove this option or derive the options from the selected model's supportedEfforts.

  { value: 'medium', labelKey: 'modelSelector.reasoningMedium', fallback: 'Medium' },

src/lib/ai/client.ts:372

• This change makes OpenRouter the provider reported by the client, but the existing isProviderErrorMessage classifier still recognizes only OpenAI, Anthropic, and Google. A backend failure such as OpenRouter error: ... has no error code and will therefore be surfaced as a generic AIError instead of AIProviderError, bypassing provider-specific handling and messaging; update the classifier/label mapping with this provider migration.

  const provider = config.provider === 'mock' ? 'mock' : 'openrouter';

supabase/functions/ai-stream/index.ts:427

• OpenRouter only sends the terminal usage chunk for a streaming request when stream_options.include_usage is enabled. With only stream: true here, streamUsageRaw will normally remain unset and successful Ask streams will be logged with cost_usd = null, so the new reported-cost accounting does not work for streaming. Add the usage option and assert it in the handler test.

• Files reviewed: 37/37 changed files
• Comments generated: 1
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.