Skip to main content

Open demo of memexa — a self-hosted Chinese personal memory graph. The full engine is the proprietary memexa product.

Project description

Memexa · 镜我

English · 中文

A self-hosted personal memory graph over Chinese-native data — WeChat / QQ / email / documents / audio. This repository is the open demo. The full engine is a separate proprietary product.

CI License: Apache 2.0 PyPI Python 3.10+

What this is

memexa turns scattered, multi-party Chinese data into a queryable memory graph: every message is stored verbatim, extracted into structured cards, and every answer is cited back to the original sentence. It is fully self-hosted — your data never leaves your machine.

This repository ships the open demo: a small synthetic dataset and a stub extractor, so you can see the shape of the project in thirty seconds — no backend, no API key, no configuration.

Try the demo

pip install memexa
memexa demo

Six synthetic sources (WeChat / QQ / email / browser / AI chat / audio) are ingested with the stub extractor, then five sample queries run against the resulting cards — entirely in memory. This is the honest first look at what the project does.

The full engine

The demo runs a stub on synthetic data. The full memexa engine is a proprietary product and is not included in this repository. It adds:

  • Live ingestion of your own data across multiple sources, incremental.
  • A two-LLM extraction pipeline producing cards with per-claim citations and cross-alias canonical identities.
  • A multi-channel recall stack with cross-encoder re-ranking — built for high-accuracy retrieval over messy, multi-party Chinese chat, not a single-vector lookup.
  • An MCP server + CLI, so any coding agent (Claude Code, Cursor, Cline, Codex) can use your memory as a first-class tool.
  • A local desktop app — runs the whole stack on your own machine.

For access to the full engine, please reach out via the repository owner's profile.

License

The demo in this repository is licensed under Apache-2.0 (see LICENSE). The full memexa engine is a separate proprietary product and is not covered by that license.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

memexa-0.1.0-1-py3-none-any.whl (20.4 kB view details)

Uploaded Python 3

File details

Details for the file memexa-0.1.0-1-py3-none-any.whl.

File metadata

  • Download URL: memexa-0.1.0-1-py3-none-any.whl
  • Upload date:
  • Size: 20.4 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for memexa-0.1.0-1-py3-none-any.whl
Algorithm Hash digest
SHA256 2ffa93b651951fc63424fa738cbf5ff7076f38fe1b1c9c5373153d300a4a0123
MD5 d6465e43b7fb022ae6eb446404aaf385
BLAKE2b-256 688833d0b40100e9affef0cbe1ee16e65bc20c32ea5011df13e64c3c58045c32

See more details on using hashes here.

Provenance

The following attestation bundles were made for memexa-0.1.0-1-py3-none-any.whl:

Publisher: publish.yml on labazhou2024/memexa

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page