blog-localize

This skill performs deep cultural adaptation of already-translated blog posts, layered above a separate blog-translate step. It solves the problem of translated content that still reads as "translated from English" — foreign brand examples, US-only statistics, mismatched formality, and irrelevant legal references — by adapting the reading experience so it feels native to the target market. It ships built-in cultural profiles for DACH, Francophone, Hispanic, and Japanese markets.

The single-file SKILL.md defines a four-phase workflow: locale understanding (parsing ISO 639-1/15924/3166 codes and loading the shared cultural profile), cultural audit (scanning for foreign-origin markers like US brands, USD pricing, and CCPA/FTC references), adaptation (substituting brand examples, localizing statistics from primary local sources, rewriting CTAs and tone per profile, mapping legal references such as CCPA to DSGVO/RGPD/LGPD, and applying Sie/du and tu/vous formality), and quality verification. It provides substitution tables and defers to a shared cultural-adaptation reference. The skill is notably security-conscious: it resolves files inside the project root, rejects symlinks, traversal, oversized (>10MB) and binary files, and gates any citation fetching behind SSRF checks (https-only, rejecting javascript/data/file schemes and localhost/private/link-local/reserved IPs after DNS resolution, disabling or revalidating redirects, and capping redirects, response size, and time).

The target users are localization teams and international content marketers using the claude-blog suite. Use cases include preparing English content for German, French, Spanish/LATAM, or Japanese audiences with locally relevant examples, statistics, and compliant legal framing. Its web access for finding local examples and statistics is read-oriented and hardened against server-side request forgery.