Hybrid search across manual pages
Search manual pages using hybrid ranking (Postgres FTS + pgvector semantic similarity merged via Reciprocal Rank Fusion). Falls back to FTS-only if embeddings are unavailable. Supports case-insensitive substring filtering by model_number (searched in applicable_models), equipment_type, manufacturer, document_id, equipment_family, and document_type. Each result includes a snippet and nested document and equipment metadata. Results include normalized_rank (score relative to top result) alongside the raw rank. low_confidence is true when the best vector cosine similarity is below 0.10 (applies in both vector and hybrid modes). Default limit: 5 results. Pass debug=true to include rel_pos (relative page position 0.0–1.0) on each result for inspecting positional ranking behavior.
Request
- Base URL: https://mcp.techmanual.ai (the document declares no server; this is the origin it was published from)
- URL: https://mcp.techmanual.ai/v1/search
- Auth: none declared
Query parameters
Headers
Response
Successful Response
Changes
No recorded changes to this operation across all 1 revision of this API.