#001pydocs-mcp1 skills50updated 2026-07-18100% of creatorskilloccupationdescriptionupdatedcomparing-retrieval-methodssoftware-developersUse when comparing retrieval pipeline configs (BM25, dense, hybrid RRF/WSI, late-interaction, LLM tree-rerank, graph expansion) on the pydocs-mcp benchmark harness, deciding which config wins, checking a change for metric regressions against a baseline, or choosing the right dataset, split, or metric for a sweep.2026-07-18