OpenClaudia Skills: 34 Marketing Skills for Claude Code

What OpenClaudia solves for people who market from a terminal

The README frames the problem as a pricing and execution gap. Most AI marketing tools, it says, charge $50 to $300 per month for a chat box that returns suggestions, and the user still has to write, publish and send. OpenClaudia takes the opposite position: the skills are free and MIT licensed, they run locally, and the README claims they 'actually write, publish, send'. The comparison table in the README lists execution, data privacy, customization and integration as the four axes where it claims an advantage.

The audience follows from that. This is for engineers, technical marketers and solo founders who already have Claude Code installed and authenticated, and who are willing to run marketing tasks as commands rather than clicking through a SaaS interface. The README gives the shape of that workflow: you open Claude Code, type a slash command such as /audit-seo https://mysite.com, and the skill runs. If you do not already work in Claude Code, nothing here meets you halfway. There is no web UI described in the README, and the homepage at openclaudia.com is listed but not documented in the repository files.

How the skills, the CLI and the API keys fit together

The repository is organized as a set of skill directories under skills/, with a cli/ directory for the installer, plus docs/, tools/ and assets/. Each skill is a self-contained unit that Claude Code loads, which is why the README can offer both a bulk install and a single-skill install. The manual path in the README is a plain copy: cp -r skills/seo-audit ~/.claude/skills/ for a personal skill, or .claude/skills/ for a project-level one. That tells you the deployment model is files on disk, not a service.

Where the design gets interesting is the API layer. Several skills are thin wrappers over third-party data sources, and the README marks which ones need credentials. semrush-research needs SEMRUSH_API_KEY, brand-monitor and brand-research need BRANDDEV_API_KEY, youtube-analytics needs YOUTUBE_API_KEY, google-reviews needs DATAFORSEO_LOGIN, and the Enception-backed skills need ENCEPTION_API_KEY or ENCEPTION_ANALYSIS_API_KEY. Google Analytics, Search Console and Google Ads reporting go through Google OAuth instead. A few skills need nothing: similarweb-traffic is listed with no API requirement, and github-stars uses the gh CLI.

That split is the real architecture. The reasoning, the copy generation and the orchestration happen inside Claude Code on your machine. The factual inputs, rankings, traffic estimates, review counts, brand data, come from paid or OAuth-gated third parties. The README's privacy claim ('runs locally on your machine') is accurate about the skill code, but it should not be read as a claim that no data leaves the machine. Any skill that calls SemRush or Brand.dev is sending a query to that vendor.

Installing OpenClaudia and running a first SEO audit

The README lists one prerequisite: Claude Code installed and authenticated. The primary install path is the npm package, invoked with npx, which installs every skill in one command.

npx openclaudia install --all

If you would rather not install the full set, the README shows a targeted form. This installs only the three named skills, which is the better starting point if you want to see what a skill does before committing to the whole collection.

npx openclaudia install seo-audit write-blog email-sequence

There is a second installer path through the skills CLI, which can also take a single skill by name. The README presents this as an alternative to the npm package, not as a replacement.

npx skills add OpenClaudia/openclaudia-skills --skill seo-audit

Once installed, the workflow moves into Claude Code itself. The README's example session shows the slash commands and the kind of output the documentation expects, including a score and an issue count for the audit.

$ claude

> /audit-seo https://mysite.com
> /write-blog "10 Tips for SaaS Growth"
> /email-sequence --type product-launch

Before you rely on any of the reporting skills, check the README's API table for the credential that skill needs and set it in your environment. A skill like google-analytics will not return data without the Google OAuth flow completed, and the README does not describe that flow in the section reproduced here.

Where OpenClaudia breaks down

The most concrete limitation is that it is not a standalone product. Claude Code is a hard prerequisite, so OpenClaudia is an extension of someone else's tool and inherits its availability, its authentication model and its pricing. If your team is standardized on a different assistant or a browser-based workflow, this collection does not help you.

The second limitation is credential sprawl. A meaningful share of the useful skills, the ones that pull rankings, traffic, reviews and ad performance, require a paid API key or an OAuth grant. The README does not document rate limits, quotas, per-call costs or what happens when a key expires mid-run. You are assembling a stack of vendors, and each one has its own billing and its own failure modes. Skills that need no key are the exception, not the rule.

The third issue is documentation depth. The README is a catalogue with a quick start and a comparison table. It does not document rollback, how to uninstall a single skill, how skill versioning works, or what happens when two installed skills overlap. USAGE.md is linked but its contents are not part of what is published alongside the README. For a tool whose value proposition includes 'fork, modify, extend any skill', the absence of a documented upgrade path is a gap worth noting before you build a workflow on top of it.

Finally, there is a count discrepancy. The repository description says 34 skills. The README says 75 in the heading and in the install comment. Both numbers appear in the project's own materials, and the README is truncated, so the true figure cannot be settled from what is available. Treat any specific number with caution.

OpenClaudia compared with hosted marketing SaaS

The obvious alternative is the category the README positions against: subscription marketing platforms that bundle SEO tracking, email sending and social scheduling behind a web interface. The difference in approach is not just price. A hosted platform owns the data pipeline, the scheduling, the retries and the reporting UI, and it takes responsibility for the integrations staying up. OpenClaudia hands all of that to you. You supply the API keys, you run the command, you read the output.

That trade is worth naming plainly. With a hosted tool you get a dashboard and a support address, and you give up the ability to read and change the code that produces your reports. With OpenClaudia you can open skills/seo-audit and edit the prompts or the logic, which the README explicitly invites. You also get the failure modes: a missing environment variable, an expired OAuth token, a vendor API change. For an engineer, the second set of problems is usually more tolerable than the first. For a marketing team without anyone who can debug a Node CLI, it is not.

A narrower alternative for the same job is writing your own Claude Code skills. The manual install path in the README, copying a directory into ~/.claude/skills/, is the same mechanism you would use for a skill you wrote yourself. OpenClaudia's value is that the prompts and the API wiring already exist for common marketing tasks, not that the mechanism is novel.

Maintenance, licensing and the cost of keeping it running

The licence is MIT, stated in the README badge and in the LICENSE file at the repository root. MIT is permissive: you can use, modify and redistribute the skills, including commercially, provided the copyright notice and licence text are preserved. This is not legal advice, and the practical implication worth checking is the interaction between the MIT-licensed skill code and the third-party APIs those skills call. Each vendor, SemRush, Brand.dev, DataForSEO, Enception, Google, has its own terms governing what you may do with the data you retrieve. The MIT licence on this repository does not grant you anything from those vendors.

The maintenance picture is less clear. The repository is not archived, but no last push date and no releases are published in the repository metadata. That means there is no basis for describing the project as actively maintained or regularly updated. What can be said is that there is a CONTRIBUTING.md, an issues tracker and a docs/ directory, which indicates the project is structured for outside contribution. Whether it is receiving that contribution is a separate question, and one you should answer by looking at the commit history on the default branch before you depend on a skill in production.

Upgrade cost is the part the README leaves open. Because skills are copied into ~/.claude/skills/ or .claude/skills/, an update means re-running the installer or re-copying the directory. If you have edited a skill locally, you need to know whether that edit survives. The README does not say. For a small number of skills this is a manual diff. Across a large installed set, it is the kind of thing that quietly breaks a scheduled workflow.

Editorial conclusion

Adopt OpenClaudia if your marketing work already happens next to Claude Code and you are comfortable supplying your own API keys for SemRush, Brand.dev, YouTube or Google OAuth. Do not adopt it if you need a hosted dashboard, a support contract, or a tool that works without an authenticated Claude Code install, because the README lists Claude Code as a prerequisite. Before committing, verify three things: which skills you actually need, whether you can obtain the API keys those skills require, and how many skills the repository currently ships, since the description says 34 while the README says 75.

Frequently asked questions

Official sources

1. Official documentation
2. Official README
3. Project repository