docs: update InvokeLLM model options to match backend

[wixysam]

Summary

• Adds gpt_5_4 to the InvokeLLMParams.model type and JSDoc options list
• Renames gemini_3_pro → gemini_3_1_pro to match the backend RuntimeModel enum
• Documents the node_modules symlink approach for running create-docs from a git worktree

Test plan

• Verify npm run create-docs regenerates correctly
• Check generated integrations.mdx reflects the updated model list

🤖 Generated with Claude Code

[claude]

Claude finished @wixysam's task —— View job

---

Pull Request Review

✅ Overall Assessment

The changes in this PR are well-focused and accurately address updating the InvokeLLM model options to match the backend. The implementation is clean and follows the existing code patterns.

📝 Detailed Review

Code Quality and Best Practices ✅

src/modules/integrations.types.ts:53

• The model options update on line 53 correctly adds gpt_5_4 and renames gemini_3_pro to gemini_3_1_pro
• The changes maintain consistency with the JSDoc comment on line 51
• The type literal union pattern follows TypeScript best practices

Documentation

• The JSDoc documentation properly reflects the available model options
• The example in the JSDoc (line 26) uses a valid model option

Potential Bugs or Issues ✅

No bugs identified. The changes are straightforward type updates that:

• Maintain backward compatibility by keeping the model parameter optional
• Add valid new model options that presumably match the backend enum
• Properly rename the deprecated Gemini model

Performance Considerations ✅

No performance impact. These are TypeScript type definition changes that:

• Have no runtime overhead
• Don't affect bundle size
• Only provide compile-time type safety

Security Concerns ✅

No security concerns identified. The changes:

• Don't expose any sensitive information
• Don't modify authentication or authorization logic
• Only update type definitions for existing functionality

Test Coverage ⚠️

The existing test coverage in tests/unit/integrations.test.ts is good for the InvokeLLM function, but I notice:

• No tests specifically use the new gpt_5_4 model
• No tests use the renamed gemini_3_1_pro model
• Existing tests use gpt_5 and claude_sonnet_4_6, which validates the type system works

Recommendation: While not blocking, consider adding a test case that uses one of the newly added models to ensure the API accepts them.

🔍 Additional Observations

1. Documentation addition in .claude/skills/sdk-docs-writing/references/pipeline-config.md: The git worktree documentation is helpful for developers working with multiple branches. This is a good addition to the project's internal documentation.

2. Commit message accuracy: The PR correctly identifies this as a documentation update since it affects TypeScript types that will be reflected in the generated documentation.

✅ Conclusion

This is a clean, focused PR that successfully updates the model options to match the backend. The changes follow best practices and maintain consistency throughout the codebase. The PR is ready to merge.

[github-actions]

🚀 Package Preview Available!

---

Install this PR's preview build with npm:

npm i @base44-preview/sdk@0.8.24-pr.159.d5c24bd

Prefer not to change any import paths? Install using npm alias so your code still imports @base44/sdk:

npm i "@base44/sdk@npm:@base44-preview/sdk@0.8.24-pr.159.d5c24bd"

Or add it to your package.json dependencies:

{
  "dependencies": {
    "@base44/sdk": "npm:@base44-preview/sdk@0.8.24-pr.159.d5c24bd"
  }
}

• 📦 Preview Package: @base44-preview/sdk@0.8.24-pr.159.d5c24bd
• 🔗 View this commit on GitHub

---

_{Preview published to npm registry — try new features instantly!}