Knowledge Base (Vector Search)

Upload documents and search them with natural language queries. SaaS Maker automatically chunks your content, generates embeddings, and stores them for fast vector similarity search.

How it works

Create an index to organize documents by topic
Upload documents — content is automatically chunked and embedded
Search with natural language queries to find relevant chunks

Embedding models

Each project is locked to a single embedding model after its first index creation. Use GET /v1/indexes/models to list available models.

Model	Provider	Dimensions	Notes
`voyage-4-large`	Voyage AI	1024	Best quality, multilingual
`voyage-4`	Voyage AI	1024	General purpose
`voyage-4-lite`	Voyage AI	1024	Fast and cheap
`voyage-code-3`	Voyage AI	1024	Code-optimized
`gemini-embedding-001`	Google	3072	100+ languages
`@cf/baai/bge-base-en-v1.5`	Cloudflare	768	Fast English (free)
`@cf/baai/bge-large-en-v1.5`	Cloudflare	1024	Quality English (free)
`@cf/baai/bge-m3`	Cloudflare	1024	Multilingual (free)

API endpoints

List available models

GET /v1/indexes/models

Auth: None

Create an index

POST /v1/indexes

Auth: API Key

curl -X POST https://saasmaker-api.sarthakagrawal927.workers.dev/v1/indexes \
  -H "Content-Type: application/json" \
  -H "X-Project-Key: pk_abc123" \
  -d '{
    "name": "help-docs",
    "embedding_model": "@cf/baai/bge-base-en-v1.5"
  }'

Field	Type	Required	Description
`name`	string	Yes	Index name (unique per project)
`embedding_model`	string	On first use	Required on the first index creation to lock the project model
`external_id`	string	No	Optional external identifier

List indexes

GET /v1/indexes

Auth: API Key

Upload a document

POST /v1/indexes/:indexId/documents

Auth: API Key

curl -X POST https://saasmaker-api.sarthakagrawal927.workers.dev/v1/indexes/idx_123/documents \
  -H "Content-Type: application/json" \
  -H "X-Project-Key: pk_abc123" \
  -d '{
    "content": "SaaS Maker is a backend-as-a-service platform for SaaS apps...",
    "metadata": { "source": "docs", "page": "intro" }
  }'

Field	Type	Required	Description
`content`	string	Yes	Document text (max 100KB)
`metadata`	object	No	Arbitrary key-value metadata

Response:

{
  "id": "doc_abc",
  "index_id": "idx_123",
  "chunks_created": 4,
  "created_at": "2025-01-01T00:00:00Z"
}

List documents

GET /v1/indexes/:indexId/documents?page=1

Auth: API Key

Search

POST /v1/indexes/:indexId/search

Auth: API Key

curl -X POST https://saasmaker-api.sarthakagrawal927.workers.dev/v1/indexes/idx_123/search \
  -H "Content-Type: application/json" \
  -H "X-Project-Key: pk_abc123" \
  -d '{
    "query": "how do I collect feedback?",
    "top_k": 5
  }'

Field	Type	Required	Description
`query`	string	Yes	Natural language search query
`top_k`	number	No	Number of results (default 5, max 20)

Response:

{
  "results": [
    {
      "document_id": "doc_abc",
      "chunk_content": "Use the feedback endpoint to collect bugs and feature requests...",
      "score": 0.92,
      "metadata": { "source": "docs", "page": "feedback" }
    }
  ]
}

Delete a document

DELETE /v1/indexes/:indexId/documents/:docId

Auth: API Key

Deleting a document also deletes all its chunks.

Delete an index

DELETE /v1/indexes/:indexId

Auth: API Key