CoreMem

Deterministic memory retrieval for agents.

CoreMem stores raw memories and retrieves them with semantic search, transparent heuristics, recency scoring, and wake-up context without requiring an LLM call during retrieval.

Install

pip install coremem

Use the HybridDB backend when you want keyword plus vector retrieval:

pip install "coremem[hybrid]"

60-second quickstart

from coremem import MemoryCore
from coremem.backends.chroma import ChromaBackend

core = MemoryCore(backend=ChromaBackend(path="./memory"))

core.ingest("user", "I visited the Museum of Modern Art today")
core.ingest("assistant", "That sounds wonderful.")

results = core.search("When did I visit art museums?")

Core concepts

Raw memories are preserved verbatim.
Backends can use ChromaDB or HybridDB.
Heuristics boost keyword, temporal, recency, name, and quote signals.
Wake-up context gives agents compact situational awareness.

Useful APIs

ingest, ingest_many
search, search_enhanced
wake_up, deep_search_context
fetch, fetch_all, store