Problem. The package's client surface is not modularized per resource/subdomain. The ruled shape for L3 client families (principal, 2026-07-31, reaffirmed 2026-08-01 for the GitHub family explicitly) is per-resource targets and products with per-resource test targets — the realized Stripe family shape. Single-module convergence was ruled a defect for the mailgun family (swift-foundations/swift-mailgun#8 et al.); the same applies here.
Proposed outcome. Per-resource/subdomain targets and products (e.g. Repositories, Traffic, Stargazers, OAuth, Collaborators, Organizations, Users — exact partition follows the API's own resource seams), each with its own test target, dependency edges declared per target, package builds green with the decomposed manifest. Sequenced with (and ideally combined into) the C5 error-shape convergence in #6 so the family is reshaped once, not twice — the traversal-mapping design decision recorded on #6 governs both.
Problem. The package's client surface is not modularized per resource/subdomain. The ruled shape for L3 client families (principal, 2026-07-31, reaffirmed 2026-08-01 for the GitHub family explicitly) is per-resource targets and products with per-resource test targets — the realized Stripe family shape. Single-module convergence was ruled a defect for the mailgun family (swift-foundations/swift-mailgun#8 et al.); the same applies here.
Proposed outcome. Per-resource/subdomain targets and products (e.g. Repositories, Traffic, Stargazers, OAuth, Collaborators, Organizations, Users — exact partition follows the API's own resource seams), each with its own test target, dependency edges declared per target, package builds green with the decomposed manifest. Sequenced with (and ideally combined into) the C5 error-shape convergence in #6 so the family is reshaped once, not twice — the traversal-mapping design decision recorded on #6 governs both.