Claude Blog Content Engine
A Claude Code workflow for researching, writing, optimizing, auditing, and publishing blog content across platforms and languages.
The document discloses MCP capabilities, environment-variable names, dependencies, and several fallback paths, and says key values must not be read or logged. However, it does not adequately define user confirmation before file writes, API calls, or external-link checks; sensitive-data isolation, network data flows, and a uniform rollback plan are also incomplete. The mandatory community footer adds unsolicited promotional external effects, so points are deducted.
Command routing, execution flow, five delivery gates, diagnostic feedback, and a three-iteration limit are described clearly. Static evidence also shows unresolved inconsistencies or gaps, including the imperfectly aligned sub-skill count, duplicated step numbering, and no proof that dependencies or external services work on key paths. Some missing capabilities produce warnings rather than controlled blocking, so the score is capped below 10 under static calibration.
The audience, blog scenarios, platform detection, trigger phrases, arguments, and multilingual capabilities are relatively clear across many platforms. However, Chinese-language output quality, Chinese semantic triggering, mainland-China reachability, and dependencies on overseas APIs/MCP services lack direct evidence or explicit boundaries; several features only degrade when authentication or service access is unavailable.
The skill includes front matter, an MIT license, versioning, command references, reference documentation, templates, agents, fallback behavior, and maintenance-related file pointers. It lacks a verified publisher identity, a clearly assigned maintainer, a formal update path, and a changelog. Potential drift in version, counts, and workflow details also prevents full marks.
The skill covers a complete blog lifecycle from research and writing through validation and delivery, with defined templates, output formats, scoring, and delivery artifacts. However, the main value depends on sub-skills, agents, scripts, MCP servers, APIs, and browser tooling not demonstrated here, and the files contain no representative independently verified outputs. Under static calibration, effectiveness is capped at 7.
The materials provide auditable workflow details, thresholds, script and test-file names, citation rules, and some source attributions, including caveats for certain claims. This review did not execute anything, and no committed test results, CI coverage, or independent corroboration of the claimed SEO/AI-citation effects is shown. The score therefore uses the static ceiling of 5.
- Do not treat the SEO, AI-citation, traffic, or performance figures in the documentation as verified results; check each source, date, and scope.
- Obtain explicit authorization before file writes, web research, Google/Gemini/NotebookLM access, image or audio generation, or CMS actions, and disclose data sent, target services, and external effects.
- Provide reachable alternatives for Chinese users and verify overseas API/MCP availability on the target network; when unavailable, give a clear stop reason and manual fallback.
- Check the actual repository for consistency among sub-skill counts, scripts, tests, CI, dependency versions, and documented versions; do not assume all 30 sub-skills exist or run from this orchestrator file alone.
- Treat the community promotional footer as optional unless the user explicitly requests it, rather than forcing it into deliverables.
What does this skill do, and when should you use it?
Claude Blog is a blog-content skill defined at skills/blog/SKILL.md in the AgriciDaniel/claude-blog repository. It routes requests across 30 sub-skills and 5 specialized agents for research, briefs, outlines, writing, rewriting, SEO, schema, fact-checking, AI-citation optimization, audits, and multilingual publishing. It supports Next.js MDX, Astro, Hugo, Jekyll, WordPress, Ghost, 11ty, Gatsby, and static HTML. Its workflow combines a 100-point scoring rubric with a five-gate delivery contract aimed at Google rankings and AI citations.
The skill reads blog files, project structure, and optional BRAND.md, VOICE.md, and DISCOURSE.md context files. It detects a platform from extensions and project signals, then routes requests to the relevant /blog command. Workflows can use WebSearch, source URLs, Google APIs, NotebookLM, Gemini TTS, image-generation MCP, and local Python scripts for research, drafting, SEO checks, JSON-LD generation, fact verification, keyword-cannibalization analysis, charts or images, translation, localization, and multilingual QA. Outputs can include articles, briefs, audits, scorecards, schema markup, editorial calendars, localized versions, hreflang data, and delivery artifacts such as HTML, PDF, hero images, screenshots, and preflight reports.
- A solo blogger wants one /blog write <topic> workflow to cover research, outlining, drafting, SEO validation, and quality scoring.
- A marketing team needs hub-and-spoke topic-cluster planning and keyword-cannibalization checks across an existing content library.
- An agency needs consistent formatting and voice across Next.js MDX, WordPress, Hugo, and other supported publishing environments.
- An international content team wants /blog multilingual to combine writing, translation, cultural localization, and hreflang output.
- An editor wants /blog analyze, /blog audit, or /blog factcheck to evaluate quality, sources, statistics, and AI-citation readiness.
- A Claude Code skill builder wants a reference architecture for routing, agent dispatch, quality gates, and CI checks.
What are this skill's strengths and limitations?
- Covers the workflow from research through delivery instead of producing only a one-shot draft.
- Includes 30 sub-skills, 5 specialized agents, 12 content templates, and a 100-point quality rubric.
- Combines SEO, E-E-A-T, AI-citable formatting, FAQ schema, source verification, and freshness checks.
- Supports multi-platform, multilingual, topic-cluster, brand-voice, and localization workflows.
- MIT-licensed, with a public repository version that does not require AI Marketing Hub Pro membership.
- Its orchestration depends on Claude Code slash commands, subagents, and Bash/local scripts, so other clients require adaptation.
- Full scoring and delivery-contract execution require Python 3.11+; image, audio, Google, NotebookLM, and MCP features may require additional credentials or configuration.
- Network research, source URLs, Google APIs, and some media workflows are not offline capabilities.
- The material names several supported platforms but does not provide separate test evidence for each platform.
- The README and SKILL.md report different version and test-count details, so those public metrics should not be treated as fully consistent based on this material alone.
How do you install this skill?
Claude Code 1.0.33+ is required. In Claude Code, run: /plugin marketplace add AgriciDaniel/claude-blog, then run: /plugin install claude-blog@agricidaniel-claude-blog. Restart Claude Code after installation. The supplied material does not document a separate manual installation procedure for skills/blog/SKILL.md. The repository also requires Python 3.11+ for quality scoring, delivery-contract runners, and linting.
How do you use this skill?
Run /blog commands in Claude Code, for example: /blog strategy <niche>, /blog write <topic>, /blog analyze <file>, /blog rewrite <file>, /blog geo <file>, or /blog multilingual <topic> --languages en,de,fr. If no subcommand is supplied, the skill asks which action is needed. It detects the target platform from project files and extensions; when the platform is unknown, it defaults to standard Markdown.
How does this skill compare with similar options?
Compared with prompting Claude or ChatGPT directly, Claude Blog provides a structured pipeline for research, outlining, drafting, validation, scoring, and up to three iterations. Compared with hosted tools such as Copy.ai or Jasper, it is an MIT-licensed open-source Claude Code skill suite. Compared with building a custom workflow, it supplies ready-made routing, agents, templates, references, and delivery gates, but it depends on Claude Code and a local runtime.