MCP server that gives AI tools search and retrieval over Typesense-indexed Payload content
Pgvector Search
Vector search adapter that stores embeddings in Postgres with pgvector
Installation
pnpm add @zetesis/payload-pgvector About
Pgvector Search keeps the text of Payload documents as embeddings in Postgres and answers similarity queries against them. Documents are synced as they are saved, so search results follow edits without a manual rebuild. Embeddings come from any OpenAI-compatible endpoint, whether a hosted provider or a self-hosted gateway, so the choice of model stays with you.
Read-only tools such as an MCP server can query the same index. A full reindex builds new vectors before it replaces the old rows, so a failed run leaves the existing index in place. The database needs the pgvector extension and a dedicated schema.
Package info
- Package name
@zetesis/payload-pgvector- Latest version
0.1.4- Unpacked size
- 100 kB
- License
- MIT
- Weekly downloads
- 40
- Last publish
- Jul 6, 2026
More from Zetesis-Labs
View profileTracks token usage, cost and sessions for LLM agent runs
Adds AI agent management, streaming chat endpoints and gateway-backed model access to Payload
Keeps Payload collections synced to Typesense with keyword, semantic and hybrid search
Hierarchical taxonomy collection and relationship field for Payload collections
Builds Lexical rich text editors from feature and block lists, with typed block renderers
Similar plugins
More in SearchBuilds a separate search index of collection documents that stays in sync on save and delete
Syncs Payload collections to Typesense for typo-tolerant search with a React search box
AI modules for ecommerce and service sites: content, search, chatbot, support
Adds a keyboard-driven global search modal to the Payload admin panel
Syncs Payload collections to Algolia with draft-aware indexing and admin reindexing
Keeps Algolia search indexes in sync with Payload collection documents