Claude Code harness

MongoDB memory engine

Hybrid retrieval

Published SDKs

---

What I build

I design the memory layer, execution harness, and protocol bridges that let autonomous agents run reliably in production — not demo chatbots that forget everything after one session.

"An agent is only as powerful as its execution harness and its memory layer."

---

Featured repositories

🤖 cc10x Router-owned harness for Claude Code. Multi-agent routing, plan review, failure-stop gates, and durable session memory. Claude Code Harness Orchestration	🦁 ClawMongo MongoDB-native AI gateway and memory engine. Episodic memory, vector search, knowledge graphs, event-sourcing. MongoDB Memory Vector Search
🧠 Memongo Persistent memory infrastructure for long-running autonomous agents. Built for agents that need to remember across sessions. Memory Engineering Agents MongoDB	📚 Hybrid-Search-RAG Production hybrid retrieval: BM25 + semantic embeddings + cross-encoder reranking. RAG Hybrid Search Python
💬 whatsapp_ai Production WhatsApp + LLM automation at scale. WhatsApp LLM Automation	🧩 agent_with_memory Autonomous agents with persistent, queryable memory layers. Agents Memory LangChain

---

Published on NPM

Package	What it does
@romiluz/clawmongo	MongoDB-backed episodic memory gateway
@romiluz13/memory-mcp	Persistent memory MCP server
@romiluz13/EmbeDocs-MCP	Document embedding pipeline via MCP

---

Stack

Focus areas: Model Context Protocol · Agent harnesses · Memory engineering · Hybrid RAG · MongoDB vector search

---

Open to collaborations on agent infrastructure, memory systems, and MCP development.