Blog Rewrite Optimizer
Rework existing blog posts to improve both Google search performance and AI citation readiness.
The skill requires a read-only audit and explicit user approval before proceeding, while repository security material documents external fetching, APIs, MCP, project-root files, and credential boundaries. However, rewriting writes content, downloads or generates media, embeds video, and runs scripts without skill-local requirements for per-action confirmation, least privilege, backup, or rollback, so points are deducted.
The workflow is staged and includes a three-iteration cap with diagnostic reporting; repository CI and tests provide additional context. This was still a static review, with no executable reproduction of the key rewrite path. Cross-directory reference assumptions, optional tools, external services, and brittle rules such as current-date insertion and fixed thresholds reduce reproducibility, so the score remains below the static ceiling.
The description clearly states user intents, supported blog formats, trigger phrases, and the boundary with blog-geo. It does not sufficiently define non-fit cases, input constraints, or fallback behavior. Chinese-language support is not stated, and research, media, video, and some generation capabilities depend on overseas services such as Google, YouTube, Pixabay, and Unsplash, creating mainland-China reachability concerns.
The SKILL.md has structured frontmatter, an MIT license declaration, phased workflow, argument hint, and detailed delivery rules. README, LICENSE, SECURITY, CI, tests, and version metadata provide governance context. The selected skill itself lacks installation and dependency guidance, input/output examples, FAQs, explicit maintenance ownership, and a local changelog, while several cited shared references are not included in the selected evidence.
The skill covers auditing, research, rewriting, citations, FAQ, media, and verification, with a relatively clear expected completion summary. Its value is plausible, but rigid requirements may produce unnatural or inaccurate work, including mandatory statistics for every H2, fixed video expectations, AI-detection heuristics, and replacement of data. Static evidence does not verify directly publishable output, so the score is capped and reduced.
The README, CI workflow, and tests provide auditable evidence for some helper scripts and delivery-contract coherence. They do not establish the correctness of blog-rewrite's key outputs, the truth of external sources, or end-to-end quality, and no tests were executed in this review, so only limited credit is justified.
- Do not run curl/irm pipe installers or unreviewed generation, rendering, or fetch scripts; pin the revision and inspect dependencies first.
- Statistics, citations, SEO effects, and AI-detection percentages still require human verification; the skill's heuristics are not factual measurements.
- Google, YouTube, Pixabay, Unsplash, and related APIs may be unreachable from mainland China; prepare local sources or manual fallbacks.
- Keep a version-control snapshot or backup before rewriting so quality regressions can be rolled back.
What does this skill do, and when should you use it?
Blog Rewrite Optimizer is designed for existing MDX, Markdown, and HTML posts. It audits the draft, researches replacement evidence, rewrites the content, and verifies the result. Its workflow combines Google ranking signals, E-E-A-T, GEO/AEO, and passage-level AI citation readiness. It is one sub-skill in the larger claude-blog suite, so parts of the documented workflow depend on shared references, scripts, and agents.
Reads an existing blog file and detects its format; scores content quality, SEO, E-E-A-T, technical elements, and AI citation readiness; checks for fabricated or unsourced statistics; measures sentence-length variance, vocabulary diversity, and known AI phrases; scans for second-order structural writing patterns; checks YouTube embeds, FAQ schema, images, charts, freshness signals, heading hierarchy, and self-promotion; searches the blog directory for keyword cannibalization; researches replacement statistics and image sources; rewrites the article with answer-first sections, citation capsules, a summary box, FAQ content, information-gain markers, and freshness updates; adds or updates SVG charts, images, and video embeds when required; and runs the documented rendering, reviewer, and preflight delivery checks.
- A content team needs to refresh an existing Markdown or MDX article and replace outdated, unsupported, or potentially fabricated statistics.
- An SEO editor wants one workflow that improves both Google-oriented signals and the extractability of passages for AI citation systems.
- An agency needs repeatable checks for heading structure, paragraph length, FAQ schema, internal linking, and keyword cannibalization across client blogs.
- An author wants to preserve personal voice and first-hand experience while reducing formulaic or visibly AI-generated phrasing.
- A blog operator wants a pre-publication score, AI-content estimate, and structured delivery verification for a substantial rewrite.
What are this skill's strengths and limitations?
- Covers Google SEO, E-E-A-T, and AI citation readiness in one rewrite workflow.
- Checks concrete editorial and technical factors including source quality, AI-writing signals, video, images, FAQ schema, cannibalization, and freshness.
- Preserves the author's voice and first-hand experience while adding answer-first sections, citation capsules, and a summary box.
- Supports MDX, Markdown, and HTML input.
- The repository README describes a five-gate delivery contract with up to three iterations when review fails.
- The complete workflow references shared repository files, scripts, and agents, so the standalone SKILL.md may not be sufficient by itself.
- Researching replacement statistics, images, and videos requires network access; cannibalization checks also require access to the blog directory.
- The source does not provide an independent test result for this individual sub-skill; repository-wide test counts apply to the whole collection.
- The workflow pauses after the audit for user approval, so it is not designed as an immediate one-shot rewrite.
- Chart generation, image generation, and parts of the delivery contract depend on other suite components or optional integrations.
How do you install this skill?
The skill is installed as part of the full claude-blog public repository rather than as a standalone package. For Claude Code 1.0.33+, the README gives these commands: /plugin marketplace add AgriciDaniel/claude-blog, followed by /plugin install claude-blog@agricidaniel-claude-blog. It also documents cloning the repository and running install.sh. The source does not document a standalone installation procedure for skills/blog-rewrite/SKILL.md.
How do you use this skill?
Run /blog rewrite <file>, for example /blog rewrite posts/example.md. The workflow first reads the post and performs a read-only audit, then presents findings and a section-by-section optimization plan. The documented process waits for user approval before research and rewriting. Natural-language triggers include “rewrite blog,” “optimize blog,” and “refresh blog post.”
How does this skill compare with similar options?
Compared with a one-shot direct prompt, Blog Rewrite Optimizer provides staged auditing, research, rewriting, and verification, with explicit checks for source quality, AI-writing patterns, citation structure, and technical elements. It fits substantial editorial refreshes that need reviewable quality controls, not quick edits to a few sentences.