shiye-10pages/memoryhub — explained in plain English
Analysis updated 2026-05-18
Keep a single private, evidence backed memory of everything you've told different AI assistants, searchable across tools.
Let Claude Code or Claude Desktop recall confirmed facts and preferences about you through MCP's recall_memory.
Import ChatGPT or Claude conversation exports and approve which memories actually get stored, avoiding hallucinated entries.
| shiye-10pages/memoryhub | 0311119/free_registertool | 18597990650-lab/multi-agent-game | |
|---|---|---|---|
| Stars | 24 | 24 | 24 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 3/5 | 4/5 | 3/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Needs Python 3.11+ and a configured LLM API key for importing or recalling memories.
MemoryHub is a local personal memory system meant to collect conversations and memories scattered across different AI tools into one private, searchable memory library that any AI can recall from later through the Model Context Protocol, known as MCP. The dashboard and its SQLite storage run only on your own machine at 127.0.0.1, with no account and no cloud sync for browsing, keyword search, or backups. However, steps that require understanding text, such as purifying raw conversation into memory entries, generating embeddings, and semantic recall, do send that content to a cloud language model you configure yourself, with Alibaba Cloud's Bailian service as the default. Every stored memory must carry verbatim evidence from the original text, and the project treats entries with no such evidence as not allowed into the database, as a safeguard against fabricated memories. Anything the AI only infers is nominated as a candidate, and only added after a person manually approves it. The tool can import from several sources: reading Claude Code's local transcripts directly, official data exports from Claude's web and desktop apps, Claude's own cloud memory export file, ChatGPT's official export, and Codex CLI's local session files. Direct reading of desktop app caches is not supported, because those apps either encrypt local data or keep conversations in the cloud rather than locally. A background sync feature can pull in new local conversations automatically on a schedule or on demand. To run it, a user needs Python 3.11 or newer, clones the repository, and runs setup and start scripts that open a local dashboard on port 7788. First time setup walks through configuring an API key, importing memories, and approving entries from a pending queue. Supported model providers include Alibaba Cloud Bailian, OpenAI, DeepSeek, Zhipu GLM, Kimi, and SiliconFlow, or any OpenAI compatible service, without a configured key, only browsing and keyword search of existing memories work. The project checks GitHub once a day for new versions and can update itself with a single click without touching local data. It is distributed under the PolyForm Strict License, which allows free personal and non-commercial use but prohibits commercial use and prohibits redistributing modified versions.
A local, evidence backed personal memory system that stores your AI conversations and lets any AI recall them through MCP.
Mainly Python. The stack also includes Python, SQLite, MCP.
Free to use personally and non-commercially, but commercial use is not allowed and you cannot redistribute modified versions.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
double-check against the repo, no cap.