Engrene Memory Bridge
leninejunior/engrene-memory-bridge
Local-first project memory for AI coding agents, shared through plain files in .memory-bridge
Engrene Memory Bridge is a local-first memory protocol for AI coding agents that stores project context as JSONL and Markdown files in a .memory-bridge folder, so Hermes, Claude Code, Cursor and other tools can share it.
What Engrene Memory Bridge does
Engrene Memory Bridge keeps sessions, decisions and a handoff snapshot as plain JSONL and Markdown inside .memory-bridge/ in the project, so different AI tools can pick up the same context. It retrieves relevant context with a hybrid of BM25 and what the README calls Joint Embedding Vectors rather than replaying whole chat histories, and a consolidated handoff.md summarizes goals and next steps. It has zero runtime dependencies and works offline using Node.js built-ins.
On Node.js 22.5 or newer it uses node:sqlite for FTS5, semantic and hybrid search; on Node.js 20 the CLI still works but search falls back to in-memory BM25. The init command adds .memory-bridge/ to .gitignore, and the README describes how to share memory across machines through git instead. Secret redaction runs before anything is written, optional encryption protects committed memory, and a bidirectional Obsidian vault sync turns decisions and sessions into notes with wikilinks. Short mb-* commands such as mb-claude, mb-codex and mb-hermes wrap the same shared code.
Key features
- Plain JSONL and Markdown memory in .memory-bridge/
- Hybrid BM25 and vector search with an in-memory BM25 fallback on Node.js 20
- Consolidated handoff.md snapshot of goals and next steps
- Secret redaction before writing and optional encryption
- Bidirectional Obsidian vault sync with wikilinks
- mb-* wrapper commands including mb-hermes
When to use it
- Switching between Hermes, Claude Code and Cursor on the same project without losing context
- Handing a task to a remote server or SSH container through a git pull
- Browsing decision history as a graph in Obsidian
Who it is for: Developers who move between several AI coding tools and want one local, file-based project memory.
How it fits with Hermes Agent
It supports Hermes among many agents, with an mb-hermes wrapper command, and its repository carries Hermes skill topics.
How to install Engrene Memory Bridge
These commands are copied from the project's README. Check the repository for the latest steps before you run them.
npm install -g engrene-memory-bridge
npx engrene-memory-bridge initRequirements: Node.js 20 or newer, with Node.js 22.5+ for FTS5, semantic and hybrid search
FAQ
What is Engrene Memory Bridge?
Engrene Memory Bridge is an open, local-first memory protocol for AI coding agents. It stores project memory as JSONL and Markdown files so different tools can share sessions, decisions and handoff state.
Does Engrene Memory Bridge work with Hermes Agent?
Yes. Hermes is one of the supported agents, with an mb-hermes wrapper command. The same memory also works with Claude Code, Codex, Cursor and others.
How do I install Engrene Memory Bridge?
Run npm install -g engrene-memory-bridge, then npx engrene-memory-bridge init inside your repository to initialize memory.
Similar memory for Hermes Agent
All memoryLong-term memory and project map for AI coding agents from paste-in prompts or a Hermes skill
susomejias rembricSelf-hosted MCP memory, sessions and dashboard for coding agents, in one Docker image and one SQLite file
juanmackie mnemosyne-hermesHermes memory provider with local semantic search, plus a standalone SQLite store with CLI and MCP server
benben17 Hermes D1 MemoryCloudflare D1 external memory provider for Hermes Agent with FTS5 search and a durable local write queue
Trainspotting31 MemoryBuddyShared memory server on Cloudflare Workers that Hermes and other MCP clients read and write
stepanov1975 Hermes Local KnowledgeHermes plugin that routes a local question to the skill, script, runbook or document to open first
Related guides: SOUL.md for Hermes Agent: what it is and how to write one · Run multiple Hermes agents with profiles