Blog Translate: SEO Blog Localization
Turns existing blog posts into SEO-ready multilingual editions while preserving publication structure.
The skill scopes translation and explicitly preserves code, URLs, structure, and names, while requiring review of machine-translation artifacts. However, it does not require confirmation before writing files, disclose source/result data flow or agent handoffs, or define rollback. Points are deducted for incomplete least-privilege, confirmation, transparency, and recovery controls.
The phased workflow includes input validation, same-language handling, missing-file and binary-file errors, and one retry for translator-agent timeouts. The core path still depends on the blog-translator agent and references not shown here, with no translation-specific tests, detailed diagnostics, or rollback behavior. Under static calibration, the score is conservatively capped below 10.
Audience, input format, language codes, output locations, and multilingual trigger phrases are reasonably clear, and zh-CN formatting rules are present. However, unsupported locales require an online research pass, offline and non-fit boundaries are under-specified, regional selection is incomplete, and calling values such as pt-BR ISO 639-1 codes is technically imprecise.
Frontmatter, argument hints, progressive on-demand references, phased workflow, error table, quality checklist, cross-references, MIT license, author, and version are present. The selected skill lacks standalone installation/dependency notes, changelog information, explicit maintenance ownership, and an update path; several capabilities rely on agents and files outside the assessed path.
The workflow addresses keyword localization, structure preservation, metadata, integrity checks, and artifact flagging, with a defined directly usable output layout. Still, claims such as native-quality and publication-ready have no skill-specific sample or execution evidence, human review remains necessary, and agent output may require rework. Static calibration limits this dimension to 7 or below.
The skill contains auditable rules, checks, and error conditions, but the supplied tests mainly cover generic blog analysis and delivery-contract coherence, not translation paths, locale formatting, SVG length handling, or agent timeout behavior. No independently reproducible result or third-party execution evidence is shown, so only limited credit is justified.
- No user confirmation, overwrite-conflict policy, or recovery procedure is specified before writing translation files.
- The blog-translator agent, referenced files, and Task dispatch are critical runtime dependencies whose availability and permissions are not verified in this skill.
- Custom-locale research requires web searches and may depend on overseas connectivity; mainland-China or offline usability needs separate confirmation.
- Human review remains advisable for SEO term selection, legal and cultural substitutions, JSON-LD semantics, and SVG text overflow.
What does this skill do, and when should you use it?
Blog Translate is an invokable skill in the claude-blog suite for translating Markdown, MDX, or HTML posts into one or more target languages. It preserves frontmatter, markdown and HTML structure, schema JSON-LD, images, charts, links, and citation capsules while localizing keywords, metadata, numbers, dates, currencies, and quote styles. It also scans for machine-translation artifacts and reports issues for review. The skill handles language conversion first; cultural adaptation is explicitly delegated to blog-localize.
Reads a Markdown, MDX, or HTML source file, detects its source language, parses comma-separated target language codes, and validates each code. It extracts translatable titles, headings, body text, image alt text, captions, SVG chart text, FAQs, citation capsules, summary boxes, CTAs, and internal-link anchor text. It builds a keyword-localization map for each target language, invokes the blog-translator agent, and can run translations in parallel. It then updates locale frontmatter such as lang, translatedFrom, translatedDate, and slug, saves files under translations/{lang}/ or multilingual/{lang}/, checks structural parity, and flags idiom literalism, mixed-language sentences, source-format numbers or dates, and untranslated frontmatter.
- A solo publisher has an English Markdown article and needs German, French, or Japanese editions without rebuilding its page structure.
- A marketing team is expanding posts containing JSON-LD, SVG charts, images, FAQs, and citations into several locales.
- An SEO editor needs target-market keyword choices reflected consistently in translated titles, meta descriptions, and H2 headings.
- A multilingual publishing workflow needs language conversion completed before a separate cultural-adaptation pass.
What are this skill's strengths and limitations?
- Preserves markdown and HTML structure, frontmatter keys, link URLs, images, and schema JSON-LD.
- Localizes keywords, meta tags, numbers, dates, currencies, and quote conventions for each target locale.
- Supports multiple target languages and checks section, FAQ, image, SVG chart, and citation-capsule parity.
- Reports common machine-translation artifacts with file, line, and suggested-fix context.
- Depends on the claude-blog collection's blog-translator agent rather than being documented as a standalone translation executable.
- Requires local filesystem access to read and write source and translated files; no independent CLI, API, or skill-specific test suite is documented.
- It does not perform the later cultural-adaptation stage; blog-localize is required for that.
- The repository README emphasizes v1.9.0, while this skill's frontmatter declares version 1.9.1, so release-version alignment should be checked before pinning deployments.
How do you install this skill?
Install it as part of the public AgriciDaniel/claude-blog collection. The README supports these Claude Code steps:
- Run
/plugin marketplace add AgriciDaniel/claude-blog. - Run
/plugin install claude-blog@agricidaniel-claude-blog. - Restart Claude Code.
The skill is located at skills/blog-translate/SKILL.md. The README does not document a standalone installation command for this sub-skill.
How do you use this skill?
Run /blog translate <file> --to <comma-separated-codes>, for example /blog translate article.md --to de,fr,es. Trigger phrases also include “translate blog”, “blog translate”, “translate post”, “blog auf Deutsch”, and “blog en espanol”. If --to is omitted, the skill asks for target ISO 639-1 codes. After translation, run /blog localize <file> --locale <code> when cultural adaptation is needed.
How does this skill compare with similar options?
The source explicitly distinguishes Blog Translate from blog-localize: this skill performs language conversion, keyword localization, and format preservation, while blog-localize performs cultural deep-adaptation afterward. Compared with generic translation, it adds blog-specific SEO handling, frontmatter and schema preservation, chart-text processing, citation-capsule checks, and structural QA.