Let every agent remember conversations, preferences, and decisions. Local storage, MCP protocol, one command to deploy.
Open terminal and run:
Python 3.10+ · Claude · WorkBuddy · Cursor · Any MCP client
pip install heropen && heropen install
📋 Copy
Not just another vector database. A memory layer built for AI agents — semantic search, multi-agent isolation, time awareness, ready out of the box.
Vector semantic → Full-text → Fuzzy match. Three-layer pipeline, never forgets.
Each agent gets its own namespace. Cross-agent search optional, content never mixes.
New conversations automatically recall previous context. Come back a week later, it still picks up where you left off.
Data stays on your machine / NAS. No cloud uploads, no external services.
Vector search, full-text retrieval, fuzzy matching — all run locally. Zero API keys, zero external dependencies.
Python 3.10+, one command.
pip install heropen
Set agent name, generate config file.
heropen setup
Auto-writes MCP config, full support.
heropen install
After installation, tell your agent "remember I like coffee" and ask "what do I like to drink" — if it can answer, it's working.
The free tier is ready to go. Upgrade to Plus or Pro when you need team collaboration.
Pro is for enterprise-level users, custom pricing tailored to your needs. Email xiaoman260220@163.com to discuss deployment options.
Product development log, open and transparent. Not a changelog — real thoughts, decisions, and lessons learned during development.
Loading...