postgres-semantic-search

The skill content coherently describes PostgreSQL-based semantic/hybrid search capabilities using pgvector, HNSW/IVFFlat indexing, and optional ParadeDB and Cohere reranking. The data flows are primarily internal to PostgreSQL with optional external reranking calls, which matches the stated goal. A notable concern is the presence of inline credentials in documentation samples (for PostgreSQL and Cohere) that could encourage unsafe handling in real usage. No malicious download/execution patterns are evident, and there is no evident credential forwarding to untrusted third-party binaries or services. Overall, the footprint is benign and proportionate to its documented purpose, with a cautionary note on secure credential management in real deployments.