cellxgene-census

Quick Review

Excellent skill for accessing the CZ CELLxGENE Census database. The description clearly conveys the capability to query 61M+ cells with filtering and integration options. The SKILL.md is exceptionally well-structured with progressive complexity (exploration → small queries → large-scale → ML integration), comprehensive code examples, and clear best practices (e.g., always filter is_primary_data). Task knowledge is outstanding with concrete workflows for memory-efficient processing, PyTorch dataloaders, and scanpy integration. The skill addresses a highly specialized, compute-intensive domain where a CLI agent would struggle with API complexity, memory management, and domain-specific filtering syntax, providing significant token cost savings. Structure is clean with a logical overview, core patterns, best practices, and supplementary reference files. Minor room for improvement in making the description slightly more explicit about out-of-core capabilities.