If not already present, build a node for Upstash Vector

[mnk-nasir]

Ensure it supports metadata filtering directly within the Flowise UI, as this is a common pain point for users trying to segment their data (e.g., "search only within 'Legal' category")

[xXMrNidaXx]

+1 for Upstash Vector support! Serverless vector DB is great for Flowise Cloud use cases.

What the node should support:

1. Basic CRUD

   • Upsert vectors with metadata
   • Query with top-k
   • Delete by ID or filter

2. Metadata filtering (critical)

   {
     "filter": {
       "category": "Legal",
       "date": { "$gte": "2024-01-01" }
     }
   }

3. Namespace support
   Different namespaces for different use cases within one index.

Node config UI:

fields:
  - name: upstashUrl
    type: credential
  - name: upstashToken  
    type: credential
  - name: namespace
    type: string
  - name: topK
    type: number
    default: 5
  - name: metadataFilter
    type: json
    optional: true

Why Upstash over Pinecone for some users:

• Pay-per-request pricing (better for low-volume)
• Built-in REST API (no SDK needed)
• Regional endpoints for latency

Workaround until native support:
Use HTTP Request node:

POST https://your-index.upstash.io/query
Headers: Authorization: Bearer $token
Body: { "vector": [...], "topK": 5, "filter": {...} }

We integrate various vector DBs with Flowise at Revolution AI — Upstash is great for serverless deployments.

[xXMrNidaXx]

Upstash Vector node would be awesome! At RevolutionAI (https://revolutionai.io) we use serverless vector DBs.

Why Upstash Vector:

• Serverless (no infra)
• Pay-per-request
• Global edge
• REST API

Node implementation idea:

class UpstashVector {
  constructor(url, token) {
    this.client = new Index({ url, token });
  }
  
  async upsert(vectors) {
    return this.client.upsert(vectors);
  }
  
  async query(vector, topK = 10) {
    return this.client.query({ vector, topK });
  }
}

Config needed:

• UPSTASH_VECTOR_URL
• UPSTASH_VECTOR_TOKEN
• Embedding dimension
• Similarity metric

Use cases:

• Edge RAG applications
• Serverless chatbots
• Cost-effective small deployments

Would definitely use this!