SkillMachine
All skills

Localization & Cultural Fit Check

Report

Scan linguistic quality plus cultural and legal risk before you ship content to a new market.

Live output preview

Input Format: Input FormatOutputWatch the Output: Watch the Output

A plan is required to view this content

Choose a plan to access input format, sample outputs, and live previews.

View Plans →

About the skill

What it does

Before you move content into a new language and market, it scans two separate risks together and hands the decision-maker a single publish signal.

1. Linguistic quality — MQM 2.0 (Multidimensional Quality Metrics). Using the industry-standard error typology, it walks the source↔target text and labels every deviation with one of 7 top dimensions (Accuracy, Fluency, Terminology, Style, Locale conventions, Audience appropriateness, Design/markup) and a severity. Severity is weighted on the exponential penalty scale — neutral(0) / minor(1) / major(5) / critical(25). The MQM Overall Quality Score then computes Quality Score = (1 − APT/EWC) × 100, where APT is the total penalty and EWC is the source word count. A per-1000-word normalization lets you compare texts of different lengths.

2. Cultural & legal lens. Against the target market's Hofstede 6-D profile (power distance, individualism–collectivism, masculinity–femininity, uncertainty avoidance, long-term orientation, indulgence), it checks whether the tone and claims fit, and flags taboos (color, symbol, gesture, religion, number, humor) plus local advertising/labeling/data law (GDPR/KVKK wording, health claims, price display). These sit outside the score but enter the findings.

3. Skopos & transcreation axis. The fidelity axis is tuned by text type — literal accuracy for legal text, cultural resonance for slogans/marketing. With the ISO 18587 full vs light post-editing distinction it states which PE level and whether ISO 17100 second-eye revision the content requires.

Decision rule: ≥95 publish, 85–94 fix-then-publish, 70–84 revise, <70 or ≥1 critical → do-not-publish. A single critical (wrong meaning, legal violation, taboo) blocks regardless of the score.

When to use

  • A final check before moving any text/UI string/marketing campaign into a new language+market.
  • Weighing whether MT output is publish-ready and at which post-editing level (full vs light).
  • Auditing whether a transcreation/slogan/brand name carries reversed meaning, taboo, or pronunciation traps in the target culture.
  • Flagging local legal/advertising compliance in regulated sectors (health, finance, food, gaming).

Method / frameworks

  • MQM 2.0 — 7-dimension typology + exponential severities (0/1/5/25) + score (1−APT/EWC)×100 (themqm.org).
  • ISO 17100:2015 — translation services, TEP + revision (second-eye) requirement.
  • ISO 18587:2017 — MT post-editing; full vs light PE levels.
  • Hofstede 6-D — de Mooij's international-marketing application.
  • Skopos theory (Vermeer/Reiss) + transcreation — functional translation; fidelity axis by text type.

Fabricated laws/statistics are forbidden: when terminology or regulation needs verification, sourced web-research runs, and unverifiable claims go to gaps.

How do I use this skill?

You don't "run" a skill — after installing it you just tell the agent your task (e.g. ask for the relevant job), and the skill kicks in by itself when its description matches.

Upload the lokalizasyon-kultur-uyum-kontrol.zip you downloaded as-is — no packaging needed, the format is already correct (folder at root).

  1. Open Settings → Customize → Skills
  2. Upload → select the lokalizasyon-kultur-uyum-kontrol.zip you downloaded
  3. Claude reads SKILL.md; the name + description appear. Ready ✅

Scripts run in Anthropic's code-execution environment (sandbox) — not on your machine.