probe-deep-nesting

To test nested skill discovery, platforms must enumerate multi-level directory references and detect SKILL.md files buried at one, two, and three nesting depths. This probe validates that platforms correctly catalog nested SKILL.md entries while respecting directory boundaries.

Platform reference reading is tested up to a three-level nesting depth, demonstrated by accessing files like references/api/v2/migration-guide.md. This probe assesses whether platforms can consistently read resources at one, two, and three levels deep within a skill directory.

A skill-loading platform can catalog nested SKILL.md files as separate skills if it correctly traverses the directory tree and respects directory boundaries. This probe validates the platform's ability to discover and enumerate nested SKILL.md entries within multi-level references.

Validating multi-level reference reading involves checking if the platform can access resources at one, two, and three depths, such as references/overview.md and references/api/endpoints.md. This probe tests consistent behavior across all nesting levels to ensure proper resource enumeration.

Nested SKILL.md files require explicit frontmatter for proper platform discovery within a skill's directory tree. This probe ensures platforms enumerate and catalog these nested entries only when explicit frontmatter is present at each nesting depth.