...
Meta-skill that teaches an agent how to build a high-density, agent-readable SKILL.md for any scientific Python library (MNE, nilearn, PyMC, scanpy, AnnData, etc.). Use when the user wants to create a new opensci-style skill for a target library, audit or refactor an existing opensci skill, build a symbol index for precise function lookup, recover from hallucinated / deprecated API calls, or normalize multi-module / namespace-package layouts. Wraps the public repo HughYau/opensci-skill so the original Light / Medium / Heavy depth gates, module map, symbol-index, and reference templates remain authoritative; this local SKILL.md repairs the upstream YAML frontmatter and preserves provenance so the workflow can be imported as one coherent skill.
Автор: songlin she · GitHub