velixar-llamaindex

Velixar AI memory retriever for LlamaIndex. Gives your LlamaIndex pipelines access to persistent cognitive memory.

Install

pip install velixar-llamaindex

Usage

from velixar_llamaindex import VelixarRetriever

retriever = VelixarRetriever(
    api_key="vx_...",
    workspace_id="ws_abc123",
)

# Use standalone
nodes = retriever.retrieve("What does the user prefer?")

# Use in a query engine
from llama_index.core import VectorStoreIndex
index = VectorStoreIndex([])
query_engine = index.as_query_engine(retriever=retriever)

Features

• 4-tier hierarchical memory (pinned, session, semantic, org)
• Automatic salience scoring and memory promotion
• Cross-session persistence
• Encrypted embeddings (AES-256-GCM)

Links

• Velixar AI
• API Docs