Role: Lead developer Org: Open Science Laboratory (OpenSciLab) Distribution: Python package, on GitHub and PyPI
Summary
Manages the memory of a user’s interactions with LLMs. Users can tap the history of past conversations when prompting, select specific parts of an exchange with one model, and carry them over to another — bridging otherwise isolated LLM instances so switching between models is smooth.
The problem it answers is vendor lock-in at the level of the conversation: everything you tell an LLM stays trapped inside that one environment, so moving to a different model means re-explaining yourself from scratch. Memor keeps the history, lets the user pick out the parts that actually matter rather than replaying the whole transcript, and carries that selection into a session with a different model. Switching models becomes less of a fresh start each time.
Positioned as a more streamlined, user-facing alternative to solutions like LangChain for conversational state transfer.
It is part of the natural-language and LLM group of OpenSciLab tools, alongside ToCount and XNum.
Links
Related
PyCon US 2026 — Memor Poster — presented as a poster in Long Beach, May 2026. ToCount — Lightweight Token Estimator — sibling LLM tooling. XNum — Universal Numeral System Converter — another OpenSciLab tool in the same NLP group. Open Source Philosophy — why these tools get written and released at all.