ai-marketing-claude-code-skills: Marketing Frameworks Packaged as Agent Skills
Marketing frameworks that AI actually executes. Designed for Claude Code, but could be used for OpenClaw, etc.
At a glance
- What is it?
- A Shell-based collection of Claude Code skills that turn marketing frameworks into instructions an agent loads and follows. The README claims 23 free and 10 paid skills, quick/standard/deep execution modes, and compatibility beyond Claude Code, but the repository's licence and the paid tier's terms are the first things to verify before adopting it.
- Who is it for?
- Adopt it if you already run Claude Code and want marketing frameworks available as loadable instructions rather than documents you paste into prompts. Skip it if you need a maintained, licensed, versioned dependency, or if your work depends on the paid skills, since the repository lists a $9 premium version on an external storefront and gives no licence file.
- Can I use it commercially?
- Not without permission. GitHub finds no licence file in the repository, and without a licence all rights are reserved by default: you may read the code but not reuse it. Check the README, or ask the authors, before using it.
- Is it still maintained?
- Yes. The repository last received commits 180 days ago.
- What is it written in?
- Mainly Shell, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What problem the skill pack addresses
The README frames the problem as repetition. Traditional content is described as something you read, apply, and then half forget. An Agent Skill, in the project's framing, is read by the agent and applied every time. That is the whole pitch: instead of writing a fresh prompt for a positioning exercise or a homepage audit, you install a folder that carries the framework, the questions, and the output format.
The intended user is someone who already works inside Claude Code and also does marketing work: a founder writing their own positioning, a solo consultant producing case studies and cold outreach, a marketer who wants an audit of a landing page without opening a separate tool. The README lists 23 free skills plus 10 Pro skills across strategy, content, research, and conversion. The skills are grouped by job rather than by channel, which is a reasonable organising choice for someone who thinks in tasks.
It is not a marketing analytics product and not a content management system. It does not connect to your ad accounts or your CRM. Everything it does happens inside the agent session, which is both the appeal and the ceiling.
Skills as a folder convention, not a service
The mechanism is deliberately plain. Agent Skills are described as an open standard for packaging expertise as instructions an agent can follow. In this repository that means each skill is a directory containing a SKILL.md file. The README links to files such as positioning-basics/SKILL.md, homepage-audit/SKILL.md, and last30days/SKILL.md. There is no server, no API key baked into the install, and no build step. The primary language listed for the repository is Shell, which matches an install path built from git clone, mkdir, and cp.
Data flow runs one way. You install the folders into Claude Code's skills directory. When your prompt touches a topic a skill covers, the agent loads that skill and follows its instructions. The README states that Claude Code will automatically use these skills when you mention related topics, and each skill entry includes a "Use when" line listing trigger phrases. Positioning Basics, for example, is triggered by prompts like help me with positioning or who is this for. That trigger list is the closest thing to a routing table the project documents.
The March 2026 v3.1 note says all 19 free skills support quick, standard, and deep execution modes, so the same skill can return a fast answer or a deeper audit without extra prompting. Note the arithmetic: the badge and the introduction say 23 free skills, while the changelog note says 19 free skills gained execution modes. The README does not reconcile those two numbers, and that gap is worth checking against the CHANGELOG before you assume every free skill behaves the same way.
Installing into ~/.claude/skills
The README gives one install path for Claude Code, in three commands. Clone the repository, create the skills directory if it does not exist, and copy the contents across:
git clone https://github.com/BrianRWagner/ai-marketing-claude-code-skills.git mkdir -p ~/.claude/skills cp -r ai-marketing-claude-code-skills/* ~/.claude/skills/
The third command is the one to read carefully. It copies everything at the repository root into your global skills directory, not into a namespaced subfolder. Whatever else lives at the root (README, CHANGELOG, licence files, any scripts) lands in ~/.claude/skills/ alongside the skill folders. If you already have skills there, this is a merge, not a clean install, and there is no documented uninstall or upgrade command. Removing the pack means deleting the folders by hand.
The README also lists compatibility with OpenClaw, Copilot, and Cursor in a badge, and the repository description says the skills are designed for Claude Code but could be used elsewhere. The README's installation section, however, only documents the Claude Code path. So the cross-tool claim is asserted rather than demonstrated, and you would be testing it yourself. The homepage at brianrwagner.com is the only other pointer given for support.
The paid tier and the missing licence file
Two things complicate adoption more than any technical detail. First, the licence. The README carries an MIT badge linking to opensource.org, but the repository metadata supplied here lists the licence as unknown, and no LICENSE file is confirmed. An MIT badge in a README is a statement of intent, not the licence text itself. If you need to redistribute these skills inside a company or bundle them into a product, confirm that a LICENSE file exists at the repository root before you rely on the badge.
Second, the paid tier. The README advertises 10 Pro skills, and the Voice Extractor entry links to a $9 premium version on an external storefront (shopclawmart.com). The install command copies the repository contents, so it is unclear from the material whether any paid skill bodies are present in the clone or whether the paid versions live entirely off-repository behind that storefront. The README does not say. If your evaluation depends on the Pro set, treat the free repository as a preview and read the storefront's terms separately.
Neither point is disqualifying on its own. Both are the kind of ambiguity that turns into a procurement conversation, which is why they belong near the top of any adoption decision rather than in a footnote.
Where the skill model breaks down
The failure mode is structural, not incidental. A skill is instructions. It has no data of its own and no way to verify a claim about your business. The homepage audit skill can apply a conversion framework to copy you paste in, but it cannot see your analytics, your bounce rate, or which variant actually converted. The positioning skill can run you through a framework, and the output is only as good as the inputs you supply. An agent following a framework confidently is still an agent following a framework.
Several skills lean on external retrieval, and the README is thin on how that retrieval is wired. Last 30 Days Research is described as covering Reddit, X, and web from the last 30 days, with a claimed 7 minutes versus 2 hours of manual research. Reddit Insights is described as using semantic AI search. Neither entry explains which search backend, which credentials, or what happens when a source is unreachable. The one skill that explicitly disclaims a dependency is Social Card Generator, which the README says has no API dependency. That disclaimer implies the others may have one, but the README does not enumerate them. Budget for reading each SKILL.md before you trust its output.
There is also a maintenance question the README answers only partly. The last push date is March 2026 and the changelog is linked, so the project is active. But the release list supplied here is empty, which means there are no tagged versions to pin to. You install from the main branch and you get whatever is there that day. For a folder of prompt instructions that is survivable. For anything you want to reproduce across a team, it is a gap.
How this differs from prompt libraries and marketing SaaS
The obvious comparison is a prompt library, a Notion page or a GitHub gist full of marketing prompts you copy and paste. The difference is packaging and triggering. A prompt library requires you to remember the prompt exists, find it, and paste it. A skill sits in the agent's skills directory and is selected by topic. The README's own contrast is between content you read and instructions the agent follows. That distinction is real, and it is the reason to prefer this format over a shared document.
The second comparison is marketing software: an SEO suite, a social scheduling tool, a landing page analyser. Those tools hold data. They crawl your site, track rankings over time, and store results you can compare across weeks. This project holds none of that. It produces analysis in a session and the session ends. If your question is whether a page converts, a tool that measures it beats a skill that reasons about it. If your question is what framework to apply and how to structure the answer, the skill is the lighter option and it lives where you already work.
The honest position is that this is a framework delivery mechanism, not a marketing platform. Judge it on whether the frameworks are ones you would otherwise pay a consultant to walk you through, and whether having them load automatically changes your behaviour.
Who should install it, and what to check first
The fit is narrow and clear. You run Claude Code daily, you do marketing work yourself rather than handing it to an agency, and you want frameworks available without maintaining a prompt document. The install is three commands and the risk is low, because you are copying text files into a directory you control.
The misfit is equally clear. If you need a licensed, versioned dependency with tagged releases, this is not that. If the paid Pro skills are central to your use case, the repository is a preview, not the product. If your marketing decisions depend on measured data rather than applied frameworks, no skill in this pack substitutes for the tool that measures it.
Before installing, check three specific things in the repository itself: whether a LICENSE file exists at the root to back the MIT badge, what the CHANGELOG says about v3.1 and which skills the quick, standard, and deep modes actually cover, and whether the clone contains the 10 Pro skills or only the free set. Those three answers determine whether this is a useful addition to your skills folder or a folder you delete after a week.
Editorial conclusion
Adopt it if you already run Claude Code and want marketing frameworks available as loadable instructions rather than documents you paste into prompts. Skip it if you need a maintained, licensed, versioned dependency, or if your work depends on the paid skills, since the repository lists a $9 premium version on an external storefront and gives no licence file. Verify three things first: the MIT badge against an actual LICENSE file in the repository, the CHANGELOG entry for v3.1 and which skills it covers, and whether the copy step into ~/.claude/skills/ brings the paid listings with it or only the free set.
Community notes