The missing layer in AI tooling: sharing what your assistant already knows
It took my AI months to learn how I think, code, and ship. When a teammate joined the project, their AI started from zero — same codebase, same conventions, none of the context. So I built memshare : peer-to-peer AI memory sharing, with consent on both sides. The problem Every AI tool today treats memory as a product feature locked inside one account. Claude remembers things for you . ChatGPT…
The article explores a solution to the challenge of transferring AI memory between different tools and users. Currently, AI tools like Claude and ChatGPT remember information within their own accounts, but this context is not accessible to other tools or even to the user in a machine-readable format. This means that as you switch between AI tools, you must start from scratch and rebuild your knowledge.
To address this issue, the author developed memshare, a peer-to-peer AI memory sharing system that treats AI memory as a data type - plain JSON files owned by the user. This approach enables AI memory to be shared across different tools without relying on the memory features of each individual chat product.
memshare works by installing a command-line tool, memoing it to your AI tool of choice, and capturing memory as you work through conversations. When you want to share context with a teammate, you can preview the bundle and then export it, specifying tags and expiration. The recipient can then import the bundle, selecting which items to accept or reject. This process ensures that only consented memory is transferred.
The consent model is a key aspect of memshare, with four gates in place to protect privacy. Items are tagged during creation, and private items never leave the user's machine, even if tags match an export. PII is automatically blocked, and before any data leaves the user's machine, it is scanned for sensitive information. Recipients also have the ability to preview and individually accept or reject items before they are added to their AI memory.
memshare uses the Model Context Protocol (MCP) to work with any MCP client, such as Cursor, Claude Code, or GitHub Copilot. The architecture consists of a memory store, which is plain JSON files, and an MCP server for sharing. The memory store is human-readable, diffable, and git-friendly, using tools like grep, diff, and git for easy management.
The tool captures AI memory effectively and makes it visible through stats, showing how many memories have been captured and when. It also makes it easy to identify issues with capture by tracking captured memories versus those added manually or not captured at all.
Overall, memshare offers a solution to the problem of transferring AI memory between different tools and users, providing a human-readable and diffable format that can be easily managed using standard Unix tools. The consent model ensures that users have full control over their AI memory and can consent to sharing only what they wish.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
