Most Claude Code conversation lives in developer territory: build an app, refactor a module, debug a stack trace. The thing nobody is talking about is what Claude Code can do for the rest of consulting work, particularly sustainability work, where the same patterns repeat across clients and the cost of regenerating them every session quietly compounds.
This is a small, growing hub of Claude Code skills built for sustainability consultants, ESG communicators, and circular economy practitioners. Open source. Free to use. Designed to cut both consulting time and AI compute cost (the math behind that is in How Claude Code Skills Cut AI Energy Use).
What’s here
- Five skills, each a self-contained directory that drops into your Claude Code skills folder.
- Installable in two minutes: clone or copy a folder, set any required env vars, done.
- Built for real work, not demos. Each skill came out of consulting workflows I run weekly.
- Open on GitHub at github.com/Borjablm/claude-skills with downloadable ZIPs and full source.
The skills
Programmatic charts, stat cards, branded title images, and free-license stock photo search. Replaces AI image generation for structured visuals at roughly 97% less cost and energy. Five chart types (bar, stacked, time series, grouped, donut) plus HTML-to-image rendering. Includes a setup wizard that walks you through your brand colors and fonts.
The four charts in the Claude energy article were generated by this skill in about three seconds total. Worth installing if you publish more than a handful of articles a year that need charts, stat cards or branded social images.
End-to-end SEO content pipeline. Five-stage workflow: brief and competitor analysis, deep research (web and video), structured outline planning, adversarial writer/editor draft, polished delivery with SEO meta. Brand voice and calibration examples. Built for sustainability content but works for any domain.
Each stage runs as a sub-agent with its own isolated context, so quality stays high across the full pipeline. Worth installing if you write SEO articles regularly and want a reusable methodology that carries brand voice and structural rigour across pieces.
Comprehensive Google Search Console audit framework. Striking-distance keywords, content decay detection, cannibalization analysis, mobile/desktop gap, dead-page finder, brand vs non-brand health ratio. The seven-point methodology I use for client SEO reviews.
Surfaces opportunities the GSC UI buries: high-impression pages with no clicks, queries one push away from top 10, content that is quietly losing position. Worth installing if you have GSC access for your own site or for clients you advise on SEO.
Publish articles to WordPress via the REST API. Multi-site support so one CLI handles every brand you maintain. Featured image upload, category and tag resolution, draft or live publish. Natural follow-up to research-writing-assistant.
Removes the copy-paste step between your local files and the WordPress block editor. Handles multilingual sites and Polylang correctly. Worth installing if you publish to WordPress more than a couple of times a month.
Prepare technical field inspections for sustainability certifications, subsidy reconciliations, supplier audits, and similar regulatory site visits. Locates approved scope documents, reconciles invoices against the approved scope per period, generates a printable inspection checklist plus a gaps report. Generalised from a real public-subsidy inspection workflow.
Tested on hundreds of facturas reconciled against the approved IFO per anualidade. The methodology generalises to any regulatory inspection that involves checking actual work against an approved scope. Worth installing if you do field-based compliance work in sustainability, energy or public-subsidy contexts.
The roadmap is shaped around what comes up in client work: a CSRD reporting helper, a material flow analysis assistant, an ESG data review tool. If there’s a sustainability workflow you’d want as a skill, open an issue on GitHub or drop me a note.
How the skills combine in a real consulting workflow
The skills are designed to be used together. A typical consulting article workflow runs through several of them in sequence.
The cycle compounds. Each skill makes the next one cheaper, and the gsc-audit output feeds back into the planning stage of the next article. field-inspection-prep sits in a parallel stream for the operational side of the practice. When a client engagement involves on-site reconciliation or audit prep, the skill handles that work without rebuilding the methodology from scratch.
Which skill should you install first?
The honest answer depends on what you spend most of your time on this month.
Start with article-images. Highest single-task payoff. Replaces $5 to $10 per month of AI image generation plus the friction of regenerating chart code each time.
Start with research-writing-assistant. The brand voice and methodology bake in once and travel with every future piece.
Start with wordpress-publish. Removes copy-paste between your local files and the WordPress editor.
Start with gsc-audit. Surfaces what the GSC UI buries.
Start with field-inspection-prep. The most domain-specific of the set, with the highest single-engagement payoff.
article-images is the easiest first install. Run it once on a real task. Then add the next skill when you feel the friction it solves.
How to install
Each skill is a self-contained directory. There are two ways to bring one into your Claude Code setup:
.claude/skills/ folder or your user-level ~/.claude/skills/. Easiest if you want to track updates.
Each skill has its own README with setup steps (env vars, dependencies, brand customisation). Most need only Python 3.9+ and one or two free API keys for the parts that touch external services.
Project-scoped vs user-scoped: Drop a skill into your-project/.claude/skills/ if it’s tied to a specific project (different brand voice per client, for example). Drop it into ~/.claude/skills/ if it’s a tool you want available across all your work. You can do both, with project-level skills taking precedence.
Why these skills exist
Most consultants I know use Claude in a way that quietly wastes compute. Brand voice pasted into every chat. Methodology framework re-explained on every call. Chart Python regenerated from scratch every time an article needs a visual. The skills system in Claude Code was built for exactly this kind of repetition, but the documentation aims at developers, so the consulting-side use cases stay invisible.
I wrote about the architecture and the math in How Claude Code Skills Cut AI Energy Use and the broader sustainability-AI tension in Claude Code for research workflows in circular economy and sustainability. These five skills are the proof points behind those pieces.
They’re also useful in their own right, regardless of whether you read the articles. The image-generation savings alone, for anyone publishing more than a handful of articles a year, pay for themselves in time and cost on day one.
Frequently asked questions
What are Claude Code skills?
Skills are self-contained packages of instructions, scripts, and reference files that Claude Code loads on demand. Each skill exposes about 100 tokens of metadata at the top so Claude can scan many at once cheaply, with the full body of around 3,000 to 5,000 tokens loading only when the skill becomes relevant to the task at hand. The architecture means you can have a dozen skills available without paying the context cost for all of them on every session.
Do I need to know Python to use these?
Not for most of them. The skills are mostly markdown-driven workflows. article-images and gsc-audit have Python components but the user-facing interface is natural language. You describe what you want, Claude runs the underlying script. wordpress-publish is the same. Only field-inspection-prep involves more file-system work that benefits from comfort with the terminal.
Can I use them on existing projects?
Yes. Drop a skill folder into your-project/.claude/skills/ for project-scoped use, or ~/.claude/skills/ to make it available across all your projects. Both work. Project-scoped overrides user-scoped when both have the same skill, which makes per-client customisation straightforward.
How are these different from Claude Projects or a system prompt?
Claude Projects bundle context and instructions into a single conversation thread. System prompts inject static instructions at the top of every message. Skills are different in two ways. They load only when relevant, so you can have many without overhead. And they include executable scripts, not just text. For consulting workflows that involve generating files, running analyses, or publishing to APIs, skills do what neither projects nor system prompts can.
Why are these built around sustainability consulting?
Because that is the practice they came out of. Sustainability research happens to combine several Claude-friendly patterns: heavy synthesis across frameworks, recurring document types, and a low tolerance for fabricated citations. The skills work fine outside the sustainability domain. The architecture is general. The test cases, examples, and brand-voice defaults all reflect the practice they were built in.
License
MIT. Use, modify, and republish freely. Attribution to azvai.com is appreciated but not required.
Stay in the loop
If you want a heads-up when a new skill lands, or when the planned Sustainability Reporting Pack (CSRD prep, material flow analysis, ESG data review) is ready, drop me a note. I keep a short list and ship infrequently.
For issues, ideas, or pull requests, the GitHub repo is the right place. Stars also help others find this if you’ve found it useful.