Upgrade Smoot to ^6.19.0

[jonkafton]

What are the relevant tickets?

N/A

Description (What does it do?)

@mitodl/smoot-design Upgrade: v6.17.1 → v6.19.0

🎉 New Features since 16.17.1.

1. Multiline Input for AI Chat (#191)

• Added support for multiline text input in the AI chat component
• Enhanced Input component with multiline capabilities
• Users can now enter multi-line prompts in the AI chat interface

2. Chat Feedback Buttons (#188)

• Added thumbs up/down feedback buttons to AI chat messages
• New Tooltip component exported in main package
• Users can now rate AI responses
• Includes feedback submission API integration

🐛 Bug Fixes

1. Add CSRF token to AiChat feedback submission (Upgrade python in Dockerfile #193)

   • Security fix for feedback submission

2. Handle relative URLs for feedback endpoint (Docker build is broken #192)

   • Improved URL handling for feedback API calls

3. Math \boldsymbol is not rendered (Update dependency @types/react-infinite-scroller to v1.2.4 #184)

   • Fixed LaTeX math rendering for bold symbols

4. Disable yarn postinstall scripts (Update dependency @types/lodash to v4.14.200 #183)

   • Security hardening to prevent malicious postinstall scripts

5. Defensive styles to resist page overrides on inputs (Refactor course numbers #173)

   • More robust input styling that resists CSS conflicts from parent pages

6. Enable prompt submit if no problem sets (Update dependency @types/css-mediaquery to v0.1.4 #170)

   • Fixed edge case where prompt couldn't be submitted

📦 Dependency Updates

• zod updated to v4 (major version bump)
• vite security updates (v7.1.5 → v7.1.11)
• Various dev dependency updates (eslint-plugin-storybook, type-fest, etc.)

[ChristopherChudzicki]

Looks good.

Suggestion: Change the default chat urls in frontend.env to

NEXT_PUBLIC_LEARN_AI_RECOMMENDATION_ENDPOINT=https://api.rc.learn.mit.edu/ai/http/recommendation_agent/
NEXT_PUBLIC_LEARN_AI_SYLLABUS_ENDPOINT=https://api.rc.learn.mit.edu/ai/http/syllabus_agent/

currently they use https://api-learn-ai-qa.ol.mit.edu which gives me a 405 method not allowed for the feedback endpoints for some reason.

The api.rc.learn versions still error, but it's 403 which seems more reasonable... probably some cookie domain setting issue since I'm not running chat locally.

But, we should double check that the feedback buttons actually function properly in RC.

[github-actions]

OpenAPI Changes

Show/hide No detectable change.

Unexpected changes? Ensure your branch is up-to-date with main (consider rebasing).