37 marketing skills as prompt text, and where the count comes from
33 battle-tested marketing skills for OpenClaw agents — Powered by MyClaw.ai
At a glance
- What is it?
- A repository of Markdown skill definitions for OpenClaw agents, organized by CRO, copywriting, SEO and paid ads, with four live data connectors bolted on top.
- Who is it for?
- What this repository actually is worth is the prompt text, not the packaging. A skill here is a Markdown file that tells an agent what to look for, and the audit, form and paywall definitions encode real conversion heuristics that a marketing team would otherwise have to write out.
- Can I use it commercially?
- Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
- Is it still maintained?
- Yes. The repository last received commits 128 days ago.
- What is it written in?
- GitHub does not report a main language for this repository.
Answers come from the project's GitHub data, last synced on September 20, 2026, and from our analysis. They are not legal advice.
Editorial analysis
What a skill in this repository physically is
The tree is small enough to read in full: `LICENSE`, `SKILL.md`, `clawhub.yaml`, a `skills/` directory, a `sponsor/` directory, and five translations of the README for German, French, Russian and Simplified Chinese. There is no source code, no build step and no package manifest. The language field on GitHub is null, which is an accurate description of a repository whose payload is Markdown.
What sits in `skills/` is a set of directories, one per skill, and the README describes them in tables with two columns: the skill name and what it does. `product-marketing-context` is listed first under Foundation and is described as the thing to create once, because all 37 skills read it automatically.
That is the whole architecture. There is no orchestrator, no plugin API and no state. A skill is instructions an agent loads, and the value of the collection is the quality of those instructions rather than any machinery around them.
The marketing question this creates is worth being blunt about. Prompt files age quietly. Nothing in this repository will fail loudly if the guidance inside it goes stale, because there are no assertions and no tests. The `clawhub.yaml` at the root suggests the collection is published to a registry where other agents can fetch it, which adds a distribution path but not a verification one.
The skill count does not reconcile, in four different directions
This is the most concrete thing in the repository and it is genuinely confusing. Four numbers appear for what should be one quantity.
The GitHub description field says 33 battle-tested marketing skills. The README heading says 37 Battle-Tested Marketing Skills, a badge says 37 skills, and the sponsorship block says run all 37 of these skills. So the repository description undercounts by four relative to the README.
Then the README's own category tables do not reach 37 either. Foundation lists 1. CRO lists 6. Copy and Content lists 5. SEO lists 6. Paid Ads and Analytics is labelled 4 plus 4 connectors. Growth and Retention is labelled 3. That totals 25 named skills plus 4 connectors, or 29, and the Growth and Retention table is cut off before its rows appear.
So there are at least four candidate counts: 33, 37, 29 visible, and whatever the incomplete final table would add. The honest reading is that the headline numbers are marketing copy rather than a generated count, and that the category tables were assembled at a different time from the header.
None of this makes the skills bad. It does mean you should count the directories in `skills/` yourself if the number matters to you, and it is a fair reason to treat a claimed skill count as approximate.
The four data connectors are the part with dependencies
Everything else in the collection is self-contained prompt text. The connectors are not, and they are described with more energy than anything else in the README.
`google-ads-connect` claims to pull the Google Ads API so the agent can audit campaigns, find wasted spend, identify zero-conversion keywords and apply fixes. `search-console-connect` pulls Google Search Console to diagnose traffic drops, find quick-win keywords and detect cannibalization. `meta-ads-connect` pulls the Meta Marketing API to detect creative fatigue, triage the Learning Phase and fix audience overlap. `x-twitter-connect` is described as connecting TweetClaw to search tweets and replies, export followers, monitor keywords and draft reviewed responses.
Notice the verbs. Three of the four describe analysis, which is bounded and verifiable. The Google Ads one ends with apply fixes, which means writing to the ad account, and the Twitter one ends with draft reviewed responses, which implies a human in the loop. Those are different risk profiles and the README does not distinguish them beyond the word reviewed.
There is a comparison paragraph aimed at a competitor named Toprank, which is described as having 2.6k stars and focusing on Google and Meta ads plus SEO data. The claim is that this collection does the same and adds X and Twitter signal research plus 28 more skills. A competitor named in a README is a marketing artifact, not a technical claim, and you should treat the comparison as positioning rather than as an evaluation.
Six CRO skills and six SEO skills, described in one line each
The category tables are the actual content, and reading them is more informative than the framing around them.
CRO is six skills: page-cro for auditing a marketing page and prioritizing fixes by impact, signup-flow-cro for registration and trial activation, onboarding-cro for time-to-value and activation rates, form-cro for lead gen, checkout and contact forms, popup-cro for exit-intent modals and overlays, and paywall-upgrade-cro for in-app upgrade moments.
SEO is six more: seo-audit for technical and on-page diagnosis with a prioritized fix list, ai-seo for optimizing for AI search surfaces including ChatGPT, Perplexity, Google AI Overviews and Claude, programmatic-seo for generating many pages from templates and data, site-architecture for hierarchy, URLs and navigation, schema-markup for JSON-LD structured data, and content-strategy for calendars, topic clusters and keyword mapping.
Copy and Content is five: copywriting for homepage and landing pages, copy-editing for tightening existing copy, cold-email for B2B outreach sequences, email-sequence for drip and lifecycle mail, and social-content for LinkedIn, X and Instagram.
The `ai-seo` entry is the one that has aged most visibly. Naming specific AI answer surfaces in a skill description is a bet on which products persist, and it is a reminder that these files carry a maintenance clock whether or not anyone is maintaining them.
The README also recommends pairing seo-audit with the search-console-connect skill for a full data-backed audit, which is the one concrete usage suggestion offered anywhere in the document.
No releases, a licence field GitHub could not read, and a sponsor block
Three structural facts sit underneath the collection.
There are no releases at all. The releases list is empty, which means there is no tag history, no changelog and no way to pin a known state. The `CHANGELOG.md` that the README of the sibling repository references does not appear in this tree. For a collection whose value is text, that is the most consequential omission, because text changes are the only kind of change this project can have.
The license situation is ambiguous in the way that usually means a script lost. GitHub reports the license field as NOASSERTION, while the README carries an MIT badge linking to a `LICENSE` file that does exist in the tree. NOASSERTION here means automatic detection failed, not that terms were withheld. The file is the authority.
The sponsorship block is substantial. A banner promotes MyClaw.ai as a cloud-hosted OpenClaw with one-click setup and 24/7 uptime, and every call to action carries a UTM-tagged URL. The repository description also credits MyClaw.ai. That is a commercial operation attached to an MIT-licensed text collection, which is perfectly legal and worth knowing about before you assume the contents are unaffiliated neutral guidance.
The last push was 2026-06-02 and the repository has 1,045 stars, 381 forks and only 2 open issues. A fork count of 381 against 1,045 stars is a high ratio, consistent with people copying out one or two skill directories rather than depending on the whole repository.
Editorial conclusion
What this repository actually is worth is the prompt text, not the packaging. A skill here is a Markdown file that tells an agent what to look for, and the audit, form and paywall definitions encode real conversion heuristics that a marketing team would otherwise have to write out. The limits are equally clear: there is no executable logic, no test suite and no release history, so nothing here is versioned in a way you can depend on, and the connector skills depend on account access this repository does not provide. Read one skill file end to end before installing all of them, and install the foundation context skill first, since every other definition assumes it has been filled in.
Frequently asked questions
What are the top 5 marketing skills?
For OpenClaw agents specifically, the collection here names 37 of its own and organizes them by discipline. The conversion rate optimization group has six entries including page-cro, form-cro and paywall-upgrade-cro, and the SEO group has six more including seo-audit and programmatic-seo. There is no external ranking behind those choices; they are the categories the maintainers chose to organise around.
How many marketing skills are in this collection?
The GitHub description says 33, while the README heading, a badge and the sponsorship text all say 37. The README's own category tables name 25 skills plus 4 data connectors, and the Growth and Retention table ends before its rows appear. Count the directories under skills/ if the exact number matters to you.
What are the best marketing skills for Claude Code and AI agents?
Those skills target OpenClaw agents rather than Claude Code specifically, and there is no compatibility claim in the README for any other client. The collection is distributed through ClawHub and by copying skill directories into an agent's skills workspace, so the portability depends on whether your client reads the same directory layout.
Do the data connectors need API access to my ad accounts?
Yes. The Google Ads, Search Console and Meta connectors are described as pulling live data through those platforms' APIs, and the X connector reaches a service called TweetClaw. The Google Ads entry goes further and describes applying fixes, so treat it as an account with write access rather than a read-only feed, and note the connector skills are the only part of the collection with outside dependencies.
What licence is this collection under?
The README carries an MIT badge pointing at a LICENSE file that exists in the repository tree, which is the authoritative statement of terms. GitHub's own license field reads NOASSERTION, which reflects a failed automatic detection rather than a different license, so the file in the tree is what to rely on.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/leoyeai-openclaw-marketing-skills)