Tools
Memory Multimodal Ingest
OpenClaw multimodal memory ingest plugin backed by LanceDB and Gemini Embedding 2
Configuration Example
{
"plugins": {
"load": {
"paths": [
"/Users/yyc/.openclaw/workspace/plugins/memory-multimodal-ingest"
]
},
"entries": {
"memory-multimodal-ingest": {
"enabled": true,
"config": {
"embedding": {
"apiKey": "${GEMINI_API_KEY}",
"model": "gemini-embedding-2-preview",
"apiBase": "https://generativelanguage.googleapis.com/v1beta",
"dimensions": 3072
},
"dbPath": "/Users/yyc/.openclaw/memory/lancedb-multimodal",
"blobPath": "/Users/yyc/.openclaw/memory/blobs",
"maxInlineBytes": 8388608
}
}
}
}
}
README
# memory-multimodal-ingest
OpenClaw plugin for multimodal memory ingest backed by LanceDB and Gemini Embedding 2.
Current MVP:
- ingest image / video / audio / PDF from local paths or HTTP(S) URLs
- embed via Gemini `gemini-embedding-2-preview`
- store vectors and metadata in LanceDB
- copy original bytes into a local blob directory
- search stored multimodal memories with text queries
- CLI commands under `openclaw memory-media`
Default config shape:
```json
{
"plugins": {
"load": {
"paths": [
"/Users/yyc/.openclaw/workspace/plugins/memory-multimodal-ingest"
]
},
"entries": {
"memory-multimodal-ingest": {
"enabled": true,
"config": {
"embedding": {
"apiKey": "${GEMINI_API_KEY}",
"model": "gemini-embedding-2-preview",
"apiBase": "https://generativelanguage.googleapis.com/v1beta",
"dimensions": 3072
},
"dbPath": "/Users/yyc/.openclaw/memory/lancedb-multimodal",
"blobPath": "/Users/yyc/.openclaw/memory/blobs",
"maxInlineBytes": 8388608
}
}
}
}
}
```
tools
Comments
Sign in to leave a comment