mcp-server-pgvector
An MCP server that gives LLM agents first-class access to pgvector-backed embedding tables in PostgreSQL: similarity search, hybrid (vector + full-text) search, upserts, and HNSW/IVFFlat index management.
에이전트 강화
게시자의 자체 명령입니다. PlugMCP 엔드포인트가 아닙니다.
Host a copy with our keys and logs.
uvx mcp-server-pgvector==0.1.0Claude Code에는 로컬 명령용 설치 링크가 없습니다. 아래 명령을 복사해 터미널에서 실행합니다.
DATABASE_URL=<secret> MCP_PGVECTOR_READ_ONLY=<value> claude mcp add io.github.mittalpk-pgvector -- uvx mcp-server-pgvector==0.1.0터미널에서 실행합니다.
{ "mcpServers": { "io.github.mittalpk-pgvector": { "command": "uvx", "args": [ "mcp-server-pgvector==0.1.0" ], "env": { "DATABASE_URL": "<secret>", "MCP_PGVECTOR_READ_ONLY": "<value>" } } }}claude_desktop_config.json
원격 URL인가요, 내 컴퓨터에서 실행하는 명령인가요? 두 설치 방식의 차이
PlugMCP에서 실행
A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.
실행게시자 제공
An MCP server that gives LLM agents first-class access to pgvector-backed embedding tables in PostgreSQL: similarity search, hybrid (vector + full-text) search, upserts, and HNSW/IVFFlat index management.
Similarity search, hybrid search, and index management for pgvector-backed PostgreSQL tables
Generic Postgres MCP servers expose raw SQL or schema introspection; this one speaks pgvector specifically — nearest-neighbor search, distance metrics, and ANN index tuning are first-class tools, not something the model has to hand-write SQL for.
information_schema / pg_catalog before being
interpolated into SQL — an LLM can only ever reference identifiers that already exist. Values are
always bound parameters.{column, op, value} allowlist, not a raw SQL fragment.MCP_PGVECTOR_READ_ONLY=true to disable upsert_embedding and create_vector_index,
leaving only read/search tools available — useful when pointing the server at a production
database.MCP_PGVECTOR_COMMAND_TIMEOUT_SECONDS, default
30s) so one expensive query can't occupy a pool connection — and stall every other caller —
indefinitely. Set it to 0 to disable.uvx mcp-server-pgvector
Or with pip:
pip install mcp-server-pgvector
python -m mcp_server_pgvector
The server reads its connection string from DATABASE_URL (or PGVECTOR_DATABASE_URL):
{
"mcpServers": {
"pgvector": {
"command": "uvx",
"args": ["mcp-server-pgvector"],
"env": {
"DATABASE_URL": "postgresql://user:password@localhost:5432/mydb",
"MCP_PGVECTOR_READ_ONLY": "false",
"MCP_PGVECTOR_COMMAND_TIMEOUT_SECONDS": "30"
}
}
}
}
Covered:
pg_catalog before use) and a closed
filter-operator allowlist — no path from tool arguments to raw SQL.twine check on the result.ConnectionRefusedError/asyncpg exceptions — verified
these don't leak the DSN's credentials into error text.Known limitations, honestly:
DATABASE_URL can
do. If you need different agents to have different permissions, give them different
connection strings backed by different Postgres roles, not different server instances of this
same DSN.hybrid_search's full-text side is hardcoded to Postgres's 'english' text search
configuration; there's no parameter to change it yet.uv sync --dev
# Bring up an isolated pgvector instance for local testing
docker compose -f docker-compose.dev.yml up -d
export DATABASE_URL=postgresql://postgres:postgres@localhost:5434/postgres
uv run pytest
uv run ruff check .
uv run pyright
See CONTRIBUTING.md. See CHANGELOG.md for release history.
MIT — see LICENSE.
Similarity search, hybrid search, and index management for pgvector-backed PostgreSQL tables
{ "mcpServers": { "io.github.mittalpk-pgvector": { "command": "uvx", "args": [ "mcp-server-pgvector==0.1.0" ], "env": { "DATABASE_URL": "<secret>", "MCP_PGVECTOR_READ_ONLY": "<value>" } } }}도구
아직 도구 목록이 없습니다. 원격 서버는 첫 상태 확인 후에 도구 목록이 생깁니다.
FAQ
이 페이지에서 Cursor, Claude Desktop, Claude Code 또는 VS Code용 stdio 설정을 복사합니다. PlugMCP 프록시가 아니라 게시자의 설치 방식을 사용합니다.
게시자가 등록한 환경 변수: DATABASE_URL, MCP_PGVECTOR_READ_ONLY.
Pgvector은(는) stdio를 통해 로컬에서 실행됩니다.
마지막 확인: 2026-11-09. Similarity search, hybrid search, and index management for pgvector-backed PostgreSQL tables