llama-index-memory-itsuki
Itsuki long-term memory for LlamaIndex, as a memory
block inside the host's own Memory class.
pip install llama-index-memory-itsuki
from llama_index.memory.itsuki import itsuki_memory
memory = itsuki_memory(user_id="user_42", session_id="thread_9")
response = await agent.run("What am I working on?", memory=memory)
Set ITSUKI_API_KEY in your environment. Create one at
itsuki.app under API Keys.
A block, not a replacement
LlamaIndex has deprecated the BaseMemory generation that older third-party
integrations target — ChatMemoryBuffer, SimpleComposableMemory and friends
are on the way out. The current architecture is the Memory class composing
BaseMemoryBlock subclasses, so that is what this ships.
Short-term chat history keeps working exactly as the host intends. Itsuki
becomes the long-term half: _aget supplies durable memory for the turn, and
_aput stages messages as they flush out of short-term memory.
Compose it yourself if you want a different arrangement:
from llama_index.core.memory import Memory
from llama_index.memory.itsuki import itsuki_memory_block
memory = Memory.from_defaults(
session_id="thread_9",
memory_blocks=[itsuki_memory_block("user_42", session_id="thread_9")],
)
Async, on purpose
BaseMemoryBlock is async-first, so this uses the Itsuki SDK's
AsyncMemoryClient. A synchronous client here would block the event loop the
agent runs on, once per step, for the length of an HTTP round trip.
Scoping
| Argument | Meaning |
|---|---|
user_id |
the memory space — required, no default |
session_id |
the conversation, and the de-duplication anchor |
agent_id, run_id |
attribution only |
project_id |
attribution, and enables project-scoped recall |
search_msg_limit (default 5) controls how many recent turns inform the recall
query.
Behaviour under failure
_aget returns an empty string and _aput returns quietly. A memory outage
costs the agent context, never the answer. The API key is a private attribute,
so it never appears in model_dump(), an export, or serialized agent state.
License
Apache-2.0
1 maintainer
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file llama_index_memory_itsuki-0.1.0.tar.gz.
File metadata
- Download URL: llama_index_memory_itsuki-0.1.0.tar.gz
- Upload date:
- Size: 13.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? Yes
- Uploaded via:
twine/7.0.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
78c2f3cdc153e1dbd5d38edcb0f091edcd2e807f6931d465c1a5a5bd8ac99cc4
|
|
| MD5 |
63cec1a975007a67995f26035895e095
|
|
| BLAKE2b-256 |
d1fb950e0c1c5a2e1714beb35042969d9c09bb44fe6e83cbcc82a29e16c90cb8
|
Provenance
The following attestation bundles were made for llama_index_memory_itsuki-0.1.0.tar.gz:
Publisher:
publish-pypi.yml on 12ziyad/universal-memory-engine
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
llama_index_memory_itsuki-0.1.0.tar.gz -
Subject digest:
78c2f3cdc153e1dbd5d38edcb0f091edcd2e807f6931d465c1a5a5bd8ac99cc4 - Sigstore transparency entry: 2476873683
- Sigstore integration time:
-
Permalink:
12ziyad/universal-memory-engine@c670e3d6df2411a1fa97481daf9e625beed5c2b9 -
Branch / Tag:
refs/heads/master - Owner: https://github.com/12ziyad
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish-pypi.yml@c670e3d6df2411a1fa97481daf9e625beed5c2b9 -
Trigger Event:
workflow_dispatch
-
Statement type:
File details
Details for the file llama_index_memory_itsuki-0.1.0-py3-none-any.whl.
File metadata
- Download URL: llama_index_memory_itsuki-0.1.0-py3-none-any.whl
- Upload date:
- Size: 16.6 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? Yes
- Uploaded via:
twine/7.0.0 CPython/3.13.14
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
5996d1973ee056c620bf9d57d8e814facc85d1621db31b383eba52b5b9379512
|
|
| MD5 |
1fe12ea56970c6991939a9a814a237b2
|
|
| BLAKE2b-256 |
885a91e8d568c3777ec5138b351ddf0e4bf203d85d0a908543033aa67f0e7303
|
Provenance
The following attestation bundles were made for llama_index_memory_itsuki-0.1.0-py3-none-any.whl:
Publisher:
publish-pypi.yml on 12ziyad/universal-memory-engine
-
Statement:
-
Statement type:
https://in-toto.io/Statement/v1 -
Predicate type:
https://docs.pypi.org/attestations/publish/v1 -
Subject name:
llama_index_memory_itsuki-0.1.0-py3-none-any.whl -
Subject digest:
5996d1973ee056c620bf9d57d8e814facc85d1621db31b383eba52b5b9379512 - Sigstore transparency entry: 2476873741
- Sigstore integration time:
-
Permalink:
12ziyad/universal-memory-engine@c670e3d6df2411a1fa97481daf9e625beed5c2b9 -
Branch / Tag:
refs/heads/master - Owner: https://github.com/12ziyad
-
Access:
public
-
Token Issuer:
https://token.actions.githubusercontent.com -
Runner Environment:
github-hosted -
Publication workflow:
publish-pypi.yml@c670e3d6df2411a1fa97481daf9e625beed5c2b9 -
Trigger Event:
workflow_dispatch
-
Statement type:
