Support notebook AI access tiers and usage limits

[koutst]

Enable the notebook AI assistant for authenticated users with note access, and use the account-specific Research AI API responses to control model selection and daily usage.

• Replace the editor/moderator frontend gate while preserving note-access checks and backend access-denied handling.
• Revalidate the model catalog per user and on opening. Disable disallowed models, normalize generation options against returned capabilities, and preserve the model lock for existing conversations. Saved premium preferences fall back to an allowed model.
• Display remaining spending allowance, the model-call cap, and reset time in the user's timezone, including unlimited accounts. Refresh usage on opening, completion, limit errors, and reset.
• Track known jobs across notebook conversations and notes, disable concurrent sends, and treat server conflicts as rejected requests. Show messages only after acceptance and preserve drafts on rejection without automatically retrying.
• Handle model eligibility changes, limits reached during a run, and cancellation while the provider may still occupy the request slot.

Validation: TypeScript type-checking, ESLint on all changed files, and 17 isolated mocked-API checks passed. Checks cover downgraded model eligibility, conversation model locks, account switching, admission errors, first-message draft preservation, concurrent sends, cancellation, mid-run limits, and budget resets.

Live end-to-end validation remains pending against the updated backend. The local backend checkout still enforces the old editor/moderator gate on notebook chat routes; this frontend targets the supplied new API contract.

[vercel]

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

Project	Deployment	Actions	Updated
web	Ready	Preview	Sep 4, 2026 1:08pm UTC

[sonarqubecloud]

Quality Gate passed

Issues
5 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud