brain-db-sdk (Python)
The Python client for the Brain memory database. Speaks Brain's BRN0 wire protocol directly; no dependency on Brain's internal code.
Status: typed-graph verbs (Phase 6). The wire codec (Phase 1) is verified
byte-for-byte against the shared conformance corpus (all 86 .bin/.json
cases re-encode to identical bytes). On top of it, a synchronous connection
layer (a transport over a socket, a Connection running the handshake HELLO →
WELCOME → AUTH → AUTH_OK and request/response, a BrainClient holding the
negotiated session), the three v1 verbs with ergonomic builders (encode(),
recall() streaming to EOS / recall_frames(), forget()), a with_retry
helper + RetryPolicy (exponential backoff, server retry_after_ms), and the
typed-graph verbs: create_entity(), create_statement(), create_relation(),
upload_schema(), and materialize_procedural(); and the space/session registry
verbs: create_space() / list_spaces() / delete_space() and
create_session() / list_sessions() / delete_session() for managing the
per-request isolation unit (space) and conversation groupings (session) that a
trusted principal targets via act_as. BrainClient is built on a
MuxConnection: a background reader thread demultiplexes responses by
stream_id, so every verb is concurrency-safe and many requests run in
flight at once over one connection from multiple threads. A Pool opens a fixed
set of such connections and hands them out round-robin for socket-level
parallelism. Transparent reconnect is a later phase.
The full build plan is in the repo-root ../PLAN.md; the
layered architecture is in ../ARCHITECTURE.md.
from brain_db_sdk import (
BrainClient, EncodeBuilder, RecallBuilder, ForgetBuilder, RetryPolicy, with_retry,
)
with BrainClient.connect("127.0.0.1", 7878) as client:
stored = client.encode(EncodeBuilder("the user prefers dark mode").build())
answer = client.recall(RecallBuilder("ui preferences").limit(5).build())
# answer.answer_kind is Single/Set (fact answer.values) or Episodic
# (memory answer.results); NoSubject/NoMemory means "don't know".
# Ride out transient ResourceExhausted/Unavailable; the stable request_id
# makes the resend idempotent.
req = ForgetBuilder(stored.memory_id).build()
with_retry(lambda: client.forget(req), RetryPolicy())
Develop + verify against the corpus:
python3 -m venv .venv && .venv/bin/pip install cbor2 pytest
PYTHONPATH=src .venv/bin/python -m pytest tests/ -q
The integration suites run against a real server when BRAIN_SDK_IT_DATA is
set, and skip otherwise; scripts/it-server.sh up boots one and prints the
vars. Set BRAIN_SDK_IT_REQUIRED=1 wherever a server is meant to be reachable
so a misconfigured one fails instead of reading as green.
Layout (folder-per-concern, mirroring the reference Rust SDK):
src/brain_db_sdk/
wire/ frame + CBOR codec, opcodes, typed payloads (corpus-verified)
errors.py client error taxonomy (BrainError + subclasses)
transport.py sync read/write of whole frames over a socket
connection.py handshake + one-at-a-time request/response
client.py high-level BrainClient: connect, handshake, encode/recall/forget
verbs.py ergonomic EncodeBuilder / RecallBuilder / ForgetBuilder
retry.py RetryPolicy + with_retry (exponential backoff, server retry_after)
aio/ the asyncio tier: AsyncBrainClient, AsyncPool, async transport+mux
asyncio tier
brain_db_sdk.aio.AsyncBrainClient is the asyncio twin of BrainClient: the
same verbs with the same signatures, every one a coroutine. It speaks the same
BRN0 frames over the same multiplexed connection — a request in flight suspends
its task instead of blocking a thread, so one connection carries thousands of
concurrent requests on a single OS thread. No new dependency: it is built on
asyncio from the standard library.
import asyncio
from brain_db_sdk import Auth, EncodeBuilder, RecallBuilder
from brain_db_sdk.aio import AsyncBrainClient
async def main() -> None:
async with await AsyncBrainClient.connect(
"127.0.0.1", 18080, Auth.token(b"brain_...")
) as client:
# Independent writes cost one round trip, not three.
await asyncio.gather(
client.encode(EncodeBuilder("Ada works at Arc Labs").build()),
client.encode(EncodeBuilder("Arc Labs builds Brain").build()),
client.encode(EncodeBuilder("Brain is a memory database").build()),
)
answer = await client.recall(RecallBuilder("where does Ada work?").build())
for memory in answer.memories:
print(memory.text)
asyncio.run(main())
AsyncPool, AsyncSubscription (drained with async for) and an awaiting
with_retry come with it. Everything below the transport is shared rather than
duplicated: the frame codec, the CBOR payload types, the request builders, the
error taxonomy, the retry policy and its backoff curve, and the Auth /
ClientConfig / ConnectionInfo configuration types are the same objects both
tiers use. tests/aio_parity compares the two surfaces by reflection and fails
the build if a verb, a signature or a default ever drifts.
Two things worth knowing:
- Pick one tier per program, not per call. The sync client blocks the event
loop from inside a coroutine; wrapping the async one in
asyncio.runfrom sync code opens and tears down a loop per call and raises outright if a loop is already running (FastAPI, Jupyter). - A connection belongs to the loop it was opened on. That binding is checked on every call rather than left to chance, because a connection driven from a second loop does not raise — it waits forever on a future nothing will complete.
HTTP tier
For hosted Brain (the Arc cloud gateway) or a self-hosted brain-edge edge, the
package also ships BrainHttpClient — a JSON-over-HTTP client with the same verb
surface, field names, and error shape as the Rust and TypeScript SDKs (the
canonical contract is ../HTTP_CONTRACT.md). Use it when
you talk to Brain through an HTTP edge and authenticate with an API key; use the
wire BrainClient above for a direct socket, streaming, transactions, and
typed-graph management. It has no third-party dependency — pure urllib.
from brain_db_sdk import BrainHttpClient
brain = BrainHttpClient(api_key, base_url="https://api.arc-labs.ai")
stored = brain.encode(text="the kettle whistled")
answer = brain.recall(query="what whistled?", max_results=3)
who = brain.whoami() # namespace + space_id + permissions
Dependencies: cbor2 (runtime), pytest + pytest-asyncio (dev). CRC32C is
pure Python; the asyncio tier adds no runtime dependency.
The live HTTP edge smoke tests cover identity, capabilities, and a memory
encode/recall/list/forget lifecycle across all SDKs. Run them from the repo
root with BRAIN_SDK_IT_HTTP=https://edge.example BRAIN_SDK_IT_HTTP_KEY=brain_… scripts/edge-it.sh.
License: Apache-2.0.
Metadata
Release files for brain-db-sdk 0.4.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| brain_db_sdk-0.4.0.tar.gz | 255.8 kB | With provenance | Details for brain_db_sdk-0.4.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| brain_db_sdk-0.4.0-py3-none-any.whl | Python 3 | none | any | Details for brain_db_sdk-0.4.0-py3-none-any.whl |
Total release size: 406.7 kB
Release files / brain_db_sdk-0.4.0.tar.gz
| Download URL | brain_db_sdk-0.4.0.tar.gz |
|---|---|
| Size | 255.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
3d54121728336a36bcaf6b4fae0e2bc7ac9821debcc6521f50c5ed0537212f50
|
|
BLAKE2b-256 checksum How to use checksums |
bb90e8369dd93ebdcbd99ac3f40d488ff0cdbe5acc4ba4407c895af9d3c06fc9
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 29, 2026.
Transparency logRelease files / brain_db_sdk-0.4.0-py3-none-any.whl
| Download URL | brain_db_sdk-0.4.0-py3-none-any.whl |
|---|---|
| Size | 150.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
a1c4a962a3413ed69e259c77ee7abf29f17ccacbd650a24caef5e8fd31e515c6
|
|
BLAKE2b-256 checksum How to use checksums |
8c98edf106735b74d12bb2a22754db8c6fbeb94ec67eb3f5f93bc4294999b43c
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 29, 2026.
Transparency logSecurity for brain-db-sdk 0.4.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
Full provenance
Both files in this release have been uploaded with provenance metadata.
Attestation for brain_db_sdk-0.4.0.tar.gz
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 29, 2026.
Transparency logA note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.