feat: Anthropic Vertex AI Support

[usnavy13]

Summary

This PR adds support for running Anthropic Claude models through Google Cloud Vertex AI, addressing the feature request in #5995.

Background:

• Claude models are available on Google Cloud Vertex AI with near-identical API functionality to the direct Anthropic API
• The current implementation creates a generic Google client which doesn't support function calls, proper token limits, or streaming
• This has been a requested feature for ~9 months with community interest

Building upon:

• ✨ feat: Anthropic Vertex AI Support #9798 by @Ziyann - Initial implementation with environment variable configuration
• feat: Add Vertex AI Claude Proxy (Express.js) #10678 by @zaherweb-2 - Alternative proxy-based approach

Key improvements in this PR:

• YAML-based configuration (similar to Azure OpenAI) for better flexibility and deployment customization
• Improved error handling with specific error messages for credential issues
• Full web search support with proper header handling for Vertex AI
• Prompt caching support via automatic header filtering for Vertex AI compatibility

---

Configuration Options

Option 1: YAML Configuration (Recommended)

# librechat.yaml
endpoints:
  anthropic:
    streamRate: 20
    vertex:
      enabled: true
      region: "us-east5"  # Optional, defaults to 'us-east5'
      projectId: "${GOOGLE_PROJECT_ID}"  # Optional, auto-detected from service key
      serviceKeyFile: "/path/to/service-account.json"  # Optional

Option 2: Environment Variables

ANTHROPIC_USE_VERTEX=true
ANTHROPIC_VERTEX_REGION=us-east5
GOOGLE_SERVICE_KEY_FILE=/path/to/service-account.json

Supports all Anthropic Claude models available on Vertex AI.

Credits

• @Ziyann - Original implementation in ✨ feat: Anthropic Vertex AI Support #9798
• @aron-muon - Web search fix contribution

Closes #5995

---

Note on Documentation

If this looks like the direction you'd like to go, I'm happy to submit a corresponding documentation PR to librechat.ai. Just didn't want to get ahead of review.

---

Change Type

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update

Testing

1. Configure Vertex AI in librechat.yaml:

   endpoints:
     anthropic:
       vertex:
         enabled: true
         region: "us-east5"

2. Place Google service account key at api/data/auth.json (or configure path via serviceKeyFile)
3. Start LibreChat and select an Anthropic model

Tested scenarios:

• Text conversations with streaming
• Image input processing
• Web search functionality
• Function/tool calling via Agents

Checklist

• My code adheres to this project's style guidelines
• I have performed a self-review of my own code
• I have commented in any complex areas of my code
• I have made pertinent documentation changes
• My changes do not introduce new warnings
• I have written tests demonstrating that my changes are effective or that my feature works
• Local unit tests pass with my changes
• Any changes dependent on mine have been merged and published in downstream modules.
• A pull request for updating the documentation has been submitted.