Remy Gaskell presents Skills Atlas, a visualization and discovery layer for a company's reusable AI skills. A proposal-writing example shows how a Markdown instruction file can capture preferences and reference material once, rather than rebuilding the same context in every chat. A content-principles example extends the approach beyond checklists to reusable editorial reasoning.
Remy Gaskell demonstrates departmental grouping, dependency graphs, a searchable library and editable skill contents. Usage records show when a skill ran and who used it, while an embedded assistant helps employees find suitable skills for tasks such as design or customer support. The demonstration illustrates these features; it does not establish their effectiveness across a large organization.
Remy Gaskell describes GitHub as the shared source of truth, with skills made available through team AI tools. Team members can contribute improvements, and version history provides a way to recover earlier working instructions. Periodic usage reviews are proposed as a way to identify unused skills and distribute the practices of experienced AI users without endlessly expanding the library.
Watch on YouTube
