Skip to main content

​
Install

pip install velixar langchain

​
Usage

from velixar.integrations.langchain import VelixarMemory
from langchain.chains import ConversationChain
from langchain_openai import ChatOpenAI

memory = VelixarMemory(
    api_key="vlx_your_key",
    user_id="user_123"
)

chain = ConversationChain(
    llm=ChatOpenAI(model="gpt-4"),
    memory=memory
)

# Memories are automatically stored and recalled
response = chain.predict(input="My favorite language is Python")
response = chain.predict(input="What's my favorite language?")
# → "Your favorite language is Python"

​
How it works

VelixarMemory implements LangChain’s BaseMemory interface:
  • load_memory_variables — Searches Velixar for memories relevant to the current input
  • save_context — Stores the conversation turn as a new memory
  • clear — Deletes all memories for the user

​
Configuration

memory = VelixarMemory(
    api_key="vlx_your_key",
    user_id="user_123",
    memory_key="history",     # Variable name in prompt (default: "history")
    recall_limit=10,          # Max memories to recall per turn
    auto_tier=True,           # Auto-assign tiers based on salience
)