mcp-server-pgvector
This MCP server provides LLM agents with specialized tools for interacting with pgvector-backed PostgreSQL tables. It enables similarity search, hybrid search combining vector and full-text capabilities, upserts for embeddings, and management of HNSW/IVFFlat indexes. Unlike generic Postgres servers, it handles nearest-neighbor search and distance metrics directly without requiring the model to write raw SQL.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
What you can do with it
- Perform k-NN similarity search over vector columns with metadata filters
- Execute hybrid search blending vector similarity with full-text search
- Create and manage HNSW or IVFFlat indexes for approximate nearest neighbor search
- Discover vector tables and describe their columns and indexes
Source and maintenance
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
mcp-server-pgvector 0.1.0
{
"version": "0.1.0",
"transport": "stdio",
"identifier": "mcp-server-pgvector",
"registryType": "pypi",
"environmentVariables": [
{
"name": "DATABASE_URL",
"format": "string",
"isRequired": true,
"description": "PostgreSQL connection string, e.g. postgresql://user:password@host:5432/dbname"
},
{
"name": "MCP_PGVECTOR_READ_ONLY",
"format": "string",
"description": "Set to 'true' to disable upsert_embedding and create_vector_index"
}
]
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"mcp-server-pgvector": {
"command": "uvx",
"args": [
"mcp-server-pgvector==0.1.0"
],
"env": {
"DATABASE_URL": "<YOUR_DATABASE_URL>",
"MCP_PGVECTOR_READ_ONLY": "<YOUR_MCP_PGVECTOR_READ_ONLY>"
}
}
}
}Cline
{
"mcpServers": {
"mcp-server-pgvector": {
"command": "uvx",
"args": [
"mcp-server-pgvector==0.1.0"
],
"env": {
"DATABASE_URL": "<YOUR_DATABASE_URL>",
"MCP_PGVECTOR_READ_ONLY": "<YOUR_MCP_PGVECTOR_READ_ONLY>"
}
}
}
}VS Code
{
"servers": {
"mcp-server-pgvector": {
"type": "stdio",
"command": "uvx",
"args": [
"mcp-server-pgvector==0.1.0"
],
"env": {
"DATABASE_URL": "<YOUR_DATABASE_URL>",
"MCP_PGVECTOR_READ_ONLY": "<YOUR_MCP_PGVECTOR_READ_ONLY>"
}
}
}
}Claude Desktop
{
"mcpServers": {
"mcp-server-pgvector": {
"command": "uvx",
"args": [
"mcp-server-pgvector==0.1.0"
],
"env": {
"DATABASE_URL": "<YOUR_DATABASE_URL>",
"MCP_PGVECTOR_READ_ONLY": "<YOUR_MCP_PGVECTOR_READ_ONLY>"
}
}
}
}Let your agent find its tools
Find servers like mcp-server-pgvector from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.