install_dense_ann
Install an approximate nearest-neighbor index on a table’s embedding column to enable vector similarity search, defaulting to HNSW with dense f32 cosine.
Instructions
Install ANN on a table (direct/local only). Default algorithm=hnsw, quantization=dense (full f32 cosine). Supported pairs: hnsw×{dense,binary_sign,product}, diskann×dense, ivf×dense. product requires product_num_subvectors (must divide dimension). Set rebuild=true to drop and recreate.
Tool call template
Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {
"table": "<table>"
},
"name": "install_dense_ann"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| table | Yes | ||
| rebuild | No | Drop existing ANN on the embedding column and recreate with the requested options | |
| algorithm | No | ANN graph/structure algorithm | hnsw |
| dimension | No | ||
| diskann_l | No | DiskANN build search-list L | |
| diskann_r | No | DiskANN max degree R | |
| ivf_nlist | No | ||
| ivf_nprobe | No | ||
| provider_id | No | ||
| product_bits | No | Bits per PQ subvector (only 8 supported) | |
| quantization | No | dense = full f32 cosine; binary_sign = legacy compact Hamming; product = PQ codes | dense |
| backfill_limit | No | ||
| embedding_column | No | embedding | |
| diskann_beam_width | No | ||
| source_text_column | No | ||
| product_num_subvectors | No | Required for product quantization; must evenly divide dimension |