Vol. 001 Jeffs Brain·Documentation

Source start/python

Python

Install the Python SDK and run the reference CLI.

Install

pip install jeffs-brain-memory
uv add jeffs-brain-memory

Requires Python >=3.11.

CLI quick start

memory init ./my-brain
memory ingest ./docs/hedgehogs.md --brain ./my-brain
memory search "where do hedgehogs live?" --brain ./my-brain
memory serve --addr 127.0.0.1:18842
memory ask --brain default --question "what did we decide?"
memory remember --brain default --path memory/notes.md --content "..."
memory recall --brain default --query "auth decision"
memory reflect --brain default
memory consolidate --brain default
memory create-brain --brain eval
memory list-brains

Same subcommands as the TypeScript and Go CLIs. See the CLI reference.

Environment variables

VariablePurpose
JB_HOMEStorage root (default ~/.jeffs-brain/).
JB_TOKENHosted bearer token. When set the SDK drives the platform HttpStore.
JB_ENDPOINTHosted endpoint URL when JB_TOKEN is set.
JB_LLM_PROVIDERopenai, anthropic, ollama, or fake.
JB_LLM_MODELModel id for the chosen provider.
OLLAMA_HOSTDefault http://localhost:11434.

Programmatic use

The Python SDK mirrors the TypeScript reference surface: store, query, search, retrieval, memory, knowledge, llm, rerank, ingest, and http are all exposed under jeffs_brain_memory.

from pathlib import Path
from jeffs_brain_memory.store.fs import FsStore
from jeffs_brain_memory.knowledge import Knowledge
from jeffs_brain_memory.search import SearchIndex

brain_id = "hello-world"
root = Path("./data") / brain_id
root.mkdir(parents=True, exist_ok=True)

store = FsStore(root=root)
index = SearchIndex(db_path=root / "search.db")
knowledge = Knowledge(brain_id=brain_id, store=store, index=index)

knowledge.ingest(path=Path("./docs/hedgehogs.md"))
hits = knowledge.search(query="where do hedgehogs live?", max_results=3)
for i, hit in enumerate(hits, start=1):
    print(f"{i}. [{hit.score:.3f}] {hit.path}\n   {hit.snippet}")

For the full feature list see the Python SDK README.

Next steps