dsh-wsl-vecmem: A Local Vector Memory Plugin Based on Ollama

`dsh-wsl-vecmem` is a lightweight local semantic memory plugin for DeepSeek Harness (DSH). It generates embeddings by calling Ollama and stores them using JSON files, providing three core commands: `vecmem_status`, `vecmem_add`, and `vecmem_search`, to support text retrieval by semantic similarity. This solution requires no standalone vector database, making it suitable for maintaining a local knowledge base, but attention should be paid to WSL network configuration issues when accessing Ollama on Windows.