Bug: PGVectorRetriever crashes on initialization due to deprecated SentenceTransformer API ("list" object has no attribute "items")

[Merdan0595]

Hi team!
I discovered a reproducible bug in PGVectorRetriever that prevents using
the pgvector backend in vector-mcp.

Error

When calling create_collection (or any tool using pgvector),
the server crashes with:
AttributeError: 'list' object has no attribute 'items'

Stack trace ends in:
sentence_transformers/models/Transformer.forward
trans_features = {key: value for key, value in features.items()}
AttributeError: 'list' object has no attribute 'items'

Root cause

In vector_mcp/retriever/pgvector_retriever.py, inside __init__:

self.embed_dim = len(
    self.embedding_function(["The quick brown fox jumps over the lazy dog."])[0]
)

Here self.embedding_function is a SentenceTransformer instance.
The forward pass expects a dict of features, not a list.

Because of this, initialization of PGVectorRetriever fails.