LangChain Local — Offline AI Chains with Ollama

Run LangChain pipelines locally using Ollama and phi4-mini (or any local model). No API keys, no cloud, fully private and offline.

Requirements

• Ollama installed and running (ollama serve)
• phi4-mini pulled (ollama pull phi4-mini)
• Python packages: pip install langchain langchain-ollama langchain-community

Modes

• coding — Python/Django code generation (low temperature, precise)
• devops — Linux/Nginx/Docker shell commands
• chat — General conversation
• rag — Document-grounded answers (context-aware)

Usage Examples

• "Ask langchain-local to write a Django REST API view"
• "Use langchain-local in devops mode to show how to check disk usage"
• "Run langchain-local chat mode: explain what is RAG"
• "Use langchain-local coding mode to write a PostgreSQL backup script"

Implementation

Execute: python3 ~/.openclaw/skills/langchain-local/main.py

Parameters

• mode: coding | devops | chat | rag (default: coding)
• query: Your question or instruction

Author

Manoj — https://github.com/manoj