2 ms·
I built this because I was frustrated with how siloed AI memory is in coding agents. Claude Code's memory only works in Claude Code. Cursor's context doesn't fo
by memel06 8mo ago
I built this because I was frustrated with how siloed AI memory is in coding agents. Claude Code's memory only works in Claude Code. Cursor's context doesn't follow you elsewhere so you switch tools and lose everything. And none of it is shareable across a team easily.
Mnemonio is a provider-agnostic memory layer: structured markdown files with YAML frontmatter, a manifest index, and some LLM-powered operations: "semantic" search (just llm search over text), auto-extraction from conversations, and distillation (merge duplicates, prune stale entries). Ships as a local MCP server, a CLI, and TypeScript library, so it works wherever you work.
- Project is very early (v0.1.7): API is stable but there will be rough edges.
This is my first open source project. I'm learning as I go, both with the library itself and with shipping in public. Feedback, criticism, and contributions are all welcome.
Repo: https://github.com/Memel06/mnemonio https://github.com/Memel06/mnemonio
Website: https://mnemonio.com https://mnemonio.com
Context7 docs for your agents: https://context7.com/memel06/mnemonio https://context7.com/memel06/mnemonio