What does this skill do, and when should you use it?
SEO Image Gen is an image-generation extension inside the claude-seo repository, focused on OG previews, blog hero images, product photography, infographics, and other SEO assets. It calls Gemini through a nanobanana MCP server and applies aspect-ratio, resolution, and domain-mode defaults for each use case. The Banana extension must be installed separately, and image generation incurs usage costs. After generation, it provides SEO guidance for alt text, file naming, WebP conversion, ImageObject schema, and OG metadata.
It identifies an image use case from /seo image-gen commands or context, applies predefined aspect ratios, resolutions, and domain modes, then generates images through the set_aspect_ratio and gemini_generate_image MCP tools. Supported modes include og, hero, product, infographic, custom, and batch. It can load brand presets and reference documentation, then suggest alt text, SEO-friendly filenames, WebP conversion, file-size targets, ImageObject markup, and og:image tags. During audits, the seo-image-gen agent only analyzes existing OG/social images and creates a generation plan; it does not generate images automatically.
- An SEO consultant needs a 1200×630-style social preview for an article or landing page and uses `og` mode for a clean, text-friendly asset.
- A content team publishing a long-form article uses `hero` mode for a widescreen, dramatic blog image.
- An e-commerce team needs clean product photography on a white background and uses `product` mode.
- A marketing team needs a data-heavy vertical visual and uses `infographic` mode.
- A team comparing creative directions uses `batch` mode to create the default three variations while estimating the cost first.
How do you install this skill?
- Generation sends prompts to the external Gemini service and may involve user-supplied images; sensitive material should not be processed before retention, cross-border-transfer, and API-scope implications are confirmed.
- The installer, fallback script, and several referenced documents required by the skill were not verified in the supplied evidence; inspect the actual files, MCP signatures, and version compatibility before execution.
- Batch generation can incur real API charges; the supplied pricing is inconsistent, so obtain budget confirmation before running it.
- Stable access to Gemini or nanobanana MCP from mainland-China networks is not demonstrated; the core function may be unavailable in the target environment.
- Generated text, copyright, brand marks, dimensions, and WebP file size require human review; examples and prompt templates are not guarantees of SEO results.
- Shell / CLI
- Network access
- Local filesystem
- MCP Server
nanobanana MCP serverPython 3ImageMagick
Install the repository first: git clone --depth 1 https://github.com/AgriciDaniel/claude-seo.git, then run bash claude-seo/install.sh. Install the extension required by this skill with ./extensions/banana/install.sh. The source does not document a standalone folder layout for installing only this skill.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/AgriciDaniel/claude-seo.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/seo-image-gen" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Professional SaaS dashboard
- A dramatic editorial image about technical SEO
- Minimal studio photo of a ceramic coffee mug
- Modern SEO analytics illustration
In Claude Code, use commands such as /seo image-gen og "Professional SaaS dashboard", /seo image-gen hero "A dramatic editorial image about technical SEO", /seo image-gen product "Minimal studio photo of a ceramic coffee mug", or /seo image-gen batch "Modern SEO analytics illustration" 3. Before generating, verify that gemini_generate_image or set_aspect_ratio MCP tools are available; otherwise install Banana. After generation, review the saved path, crafted prompt, model settings, and SEO checklist.
What are this skill's strengths and limitations?
- Provides explicit presets for OG images, blog heroes, product photos, infographics, favicons, and Pinterest Pins.
- Configures aspect ratio, resolution, and domain mode by use case, reducing manual parameter setup.
- Covers post-generation SEO considerations including alt text, filenames, WebP, ImageObject, and OG tags.
- Supports batch variations and provides approximate cost guidance plus generation logging.
- Requires the nanobanana MCP server; generation tools are unavailable without the extension.
- Depends on Gemini image generation, so usage costs, API-key issues, rate limits, and safety filters can affect execution.
- The repository documents broader audit limitations around heavy hydration and interaction-loaded content, but this skill has no stated image-quality test suite.
- The documentation does not provide a standalone installation process for this individual skill or demonstrate direct use outside Claude Code's command workflow.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| SEO Visual Generator this page | 44 · Not recommended | ★ 19k | 23d ago | MIT |
| Firecrawl SEO Site Analyzer | 52 · Use with care | ★ 19k | 23d ago | MIT |
| DataForSEO Live SEO Intelligence | 47 · Use with care | ★ 19k | 23d ago | MIT |
| Ahrefs SEO Intelligence Extension | 45 · Not recommended | ★ 19k | 23d ago | MIT |
| Maps Intelligence for Local SEO | 54 · Use with care | ★ 19k | 23d ago | MIT |
This is not a complete SEO auditing tool; it is the repository extension responsible for generating visual assets. Compared with the repository's seo-images analysis skill, it performs generation, while seo-images findings can guide generation requests. It is built on the standalone Claude Banana image-generation skill and reuses its Creative Director pipeline.
How did FollowSkills review this skill?
The file discloses the nanobanana MCP requirement, GOOGLE_AI_API_KEY, the external Gemini API, and the output location, and it instructs the agent to show estimated costs before generation. However, it does not require confirmation for every generation or batch operation, and it does not adequately explain prompt/image data flows, third-party retention, permission scope, or rollback. No malware, credential theft, or covert-exfiltration red line is evident.
Commands, use-case mappings, error cases, and fallback guidance provide a plausible happy path. However, the referenced installer, Python scripts, and many reference files are not included in the supplied evidence; MCP support for key parameters is version-dependent; the 4K infographic preset conflicts with 2K routing guidance; and abnormal-input and output validation are thin. The static calibration therefore limits this to a modest score.
Trigger phrases, argument forms, scenarios, image types, and aspect-ratio mappings are relatively clear. Core operation nevertheless depends on an external Gemini/nanobanana service, with no evidence of mainland-China reachability, Chinese-language support, or availability in the target environment. Non-fit scenarios, brand/copyright constraints, and boundaries beyond image generation are also under-specified, so points are deducted.
The skill uses layered documentation covering prerequisites, quick reference, workflow, errors, and on-demand references, and declares MIT licensing, author, and version 2.2.0. It lacks a skill-specific changelog, clearly assigned maintenance responsibility, and a concrete update path; dependency/install verification is incomplete; and several referenced scripts and documents are absent from the supplied selected-skill evidence. Points are deducted for those governance and reproducibility gaps.
Use-case-to-parameter mappings, generation workflow, post-processing advice, OG tags, and ImageObject examples address the core image-generation task and may reduce manual prompt design. Yet image quality, dimensions, text accuracy, SEO impact, and file-size targets are not verified in the supplied static material. Human review and post-processing remain necessary, and batch cost-benefit is not demonstrated, so the score stays below the static ceiling.
Model details, tool parameters, error codes, pricing tables, and external references provide limited auditability. The supplied material contains no tests, CI coverage specific to this path, or third-party execution evidence; the “2,500+ tested prompts” claim is not independently substantiated; and pricing is internally inconsistent between approximately $0.039 and approximately $0.067 per image. Points are deducted accordingly.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →