openclaw-knowledgebase-generate-alibaba-mysql

OpenClaw plugin for chunking text, generating embeddings via DashScope, and storing into Alibaba Cloud RDS MySQL vector tables.

Features

• Text chunking: fixed, paragraph, and none strategies with configurable chunk size and overlap
• Embedding: DashScope-compatible (OpenAI SDK) embedding generation
• Vector storage: Alibaba Cloud RDS MySQL with VECTOR type and HNSW cosine index
• Auto table creation: Tables are created on first write if they don't exist
• Semantic deduplication: Time-windowed duplicate detection via cosine similarity

Tools

| Tool | Description | |------|-------------| | kb_store | Store a single text (auto-chunked and embedded) | | kb_store_batch | Batch store multiple items | | kb_scan | View table statistics and recent entries | | kb_dedupe | Find and archive semantic duplicates | | kb_summarize | Retrieve entries for AI summarization | | kb_archive | Soft-delete entries by ID |

Installation

openclaw plugins install openclaw-knowledgebase-generate-alibaba-mysql

Configuration

Add to openclaw.json under plugins.entries:

{
  "openclaw-knowledgebase-generate-alibaba-mysql": {
    "enabled": true,
    "config": {
      "mysql": {
        "host": "your-rds-host",
        "port": 3306,
        "user": "your-user",
        "password": "your-password",
        "database": "your-database"
      },
      "embedding": {
        "apiKey": "${DASHSCOPE_API_KEY}",
        "model": "text-embedding-v3",
        "baseUrl": "https://dashscope.aliyuncs.com/compatible-mode/v1",
        "dimensions": 1024
      },
      "defaultTable": "kb_default",
      "chunking": {
        "strategy": "paragraph",
        "maxChunkSize": 500,
        "overlap": 100
      }
    }
  }
}

License

MIT