Aaron Marketing Skills: 120 Marketing Skills Installed as an AI Staff
120 marketing skills as an AI marketing staff — plugin, portable skills, or an 8-bot team across 7 disciplines (narrative, SEO/GEO, social, email, paid, influencer, launch) on one contract, with 8 auditor gates: TALE · CORE-EEAT · CITE · ECHO · SEND · ROAS · STAR · RAMP.
At a glance
- What is it?
- Aaron Marketing Skills packages 120 marketing skills across seven disciplines into a plugin, portable skill set, or an eight-bot team, with eight auditor gates for scoring. It is for marketing teams already working inside an agent host, and the Apache-2.0 licence keeps the Markdown portable.
- Who is it for?
- Adopt Aaron Marketing Skills if your team already runs Claude Code, Codex, Cursor, or another supported host and wants marketing workflows as versioned Markdown rather than prompt snippets. Do not adopt it if you need a hosted dashboard, a single MCP server, or a guarantee that every bot tier works without your own data.
- Can I use it commercially?
- Yes. Apache-2.0 is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 4 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on September 20, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What Aaron Marketing Skills actually installs
The repository is a marketing skill library, not a SaaS product. The README describes it as "an AI marketing staff you install, not prompt": 120 agent skills distributed as one plugin, as portable skills on 70+ hosts, or as an eight-bot AI Staff on Grok Bot and Hermes Bot Mode. The skills and commands are plain Markdown, with small Bash and Python-stdlib runtimes providing hooks, validation, scoring, registry events, connectors, and CI. The README states there is no pip step.
The intended user is a marketer or growth engineer who already works inside an agent host such as Claude Code, Codex, Cursor, OpenCode, Antigravity, Gemini CLI, Copilot CLI, OpenClaw, or Hermes. The seven disciplines are narrative, SEO/GEO, social, email, paid ads, influencer, and launch, each with 16 skills and a four-phase lifecycle. A separate protocol layer of eight skills holds shared machinery: seven truth registries plus HOT/WARM/COLD memory. That split matters. The discipline skills do the marketing work; the protocol layer is what keeps entities, claims, consent, and channel facts consistent across them.
The project is Apache-2.0 and the last push was on 2026-09-20, four days before this writing. That is recent enough that the repository is not in a maintenance gap, and it is not archived. The latest release listed is v20.1.0 from 2026-09-02.
Phase directories, framework gates, and one routing command
Each discipline is organized as a lifecycle of phase directories. Narrative runs trace, architect, land, evaluate. SEO/GEO runs survey, implement, tune, evaluate. Social runs explore, craft, host, observe. Email runs setup, engage, nurture, deliver. Paid ads runs research, orchestrate, activate, scale. Influencer runs scout, target, activate, report. Launch runs research, assemble, mobilize, prove.
Each discipline also maps to a named framework that feeds an auditor gate. Narrative uses TALE, described in the README as truth, system, and effectiveness profiles. SEO/GEO uses CORE-EEAT feeding a content-quality-auditor, and CITE feeding a domain-authority-auditor. Social uses ECHO with asset and program-maturity profiles. Email uses SEND and the email-quality-auditor, which the README abbreviates as EQS. Paid ads uses ROAS and the ad-account-auditor, or RQS. Influencer uses STAR and the creator-content-auditor, or SQS, with a separate fit-scorer producing the Suitability read. Launch uses RAMP with preflight, execution, and outcome profiles.
The gate names are the part worth pausing on. They are scoring rubrics, not external certifications, and the README does not claim any third-party endorsement of them. Treat them as the project's internal quality bar.
Routing is handled by a single command, /aaron-marketing:auto, which takes a natural-language goal. The typed topology lives in references/system-catalog.json, and a readable map is in docs/system-architecture.md. If you want to see how the pieces connect before installing, those two files are the place to start, because the README's table is a summary rather than the full graph.
Installing Aaron Marketing Skills and running a first SEO/GEO task
The install path depends on your host. For Claude Code, the README gives a two-step marketplace flow. The first command only registers the catalog, so the second one is required to actually enable skills and commands.
/plugin marketplace add aaron-he-zhu/aaron-marketing-skills
/plugin install aaron-marketing@aaronFor Codex, Cursor, OpenCode, Antigravity, Gemini CLI, Copilot CLI, OpenClaw, Hermes, and the wider set of 70+ hosts, the README uses the skills CLI instead.
npx skills add aaron-he-zhu/aaron-marketing-skillsYou can install a single skill rather than the whole set. The README gives keyword-research as the example, using the -s flag.
npx skills add aaron-he-zhu/aaron-marketing-skills -s keyword-researchIf you would rather not use a package manager at all, cloning the repository is listed as an option for any host.
git clone https://github.com/aaron-he-zhu/aaron-marketing-skillsAfter install, a first real use is to invoke a discipline command rather than the router, so you can see the phase flow. The SEO/GEO entrypoint is /aaron-marketing:seo-geo. The README states that every skill runs at Tier 1 with data you provide, so expect to supply the domain, the content, or the account context yourself. There is no bundled crawler or ad-account connection implied by the install steps.
Two installation details are easy to miss. The plugin adds nothing to /mcp; docs/mcp-catalog.json is described as a copy-paste reference only. And the repository root is the authoring source of truth, not an Agent Plugins v1 install root, so for that client you should use the Portable Lite release asset, which the README lists as aaron-marketing-skills-20.1.0-agent-plugin-v1-lite.tar.gz on the v20.1.0 release page.
Where the install story gets thin
The most concrete limitation is the one the README states plainly: every skill runs at Tier 1 with data you provide. There is no indication of a hosted data layer that fills in competitor research, ad performance, or audience data on its own. If your expectation is an agent that pulls your Search Console and Meta Ads accounts and returns finished analysis, the README does not describe that. The CONNECTORS.md file exists at the repository root, so connectors are part of the design, but the README does not document which systems they reach or what credentials they need.
The AI Staff path is narrower than the rest. The eight-bot team runs on Grok Bot and Hermes Bot Mode specifically. The install instructions differ by host: Hermes uses profile install, while Grok is described as semi-manual cards. That is a real asymmetry, and the README does not present the Grok path as a one-command setup.
The documentation hub is also incomplete in a way worth flagging. The README links to aaronmarketing.ai/docs/ and labels the hub root a placeholder. So the long-form narrative and SEO material you might expect to find there is not fully available yet, and the in-repo docs/ directory is the more reliable reference.
Finally, the gate frameworks are the project's own. CORE-EEAT, CITE, ECHO, SEND, ROAS, STAR, RAMP, and TALE are defined in the references/ directory of this repository. Adopting them means adopting this project's scoring opinions, which is fine for internal consistency and less useful if you need a rubric that other teams already recognize.
How this differs from pointing an agent at a prompt library
The obvious alternative is a prompt collection or a general-purpose agent framework where you write your own marketing prompts and keep them in a shared document. The difference is structural. A prompt library gives you text with no lifecycle, no shared state, and no scoring gate. Aaron Marketing Skills gives each discipline a four-phase directory flow and a named auditor that produces a score, and it puts seven truth registries plus HOT/WARM/COLD memory underneath all of them. That memory layer is what lets the narrative skills and the launch skills agree on the same entity and the same claims.
A second alternative is a hosted marketing automation platform with an AI assistant bolted on. Those give you execution against real ad and email accounts out of the box, which this project does not claim. What they do not give you is plain Markdown you can diff, review, and version in Git alongside the rest of your repository. The README is explicit that skills and commands are plain Markdown, and the repository carries an evals/ directory and a tests/ directory, which suggests the skills themselves are treated as testable artifacts.
A third comparison is the sibling repositories. The README notes that seo-geo-claude-skills and influencer-marketing-agent-skills are signpost repos that point here, with a sibling policy documented in docs/repo-family.md. If you find one of those older repositories first, the README's position is that this bundle supersedes them for those two disciplines.
Maintenance, versioning, and what Apache-2.0 means here
The repository is not archived and the last push was on 2026-09-20. The release cadence visible in the release list is roughly two to three weeks between tagged versions: v19.2.0 on 2026-08-10, v20.0.0 on 2026-08-20, and v20.1.0 on 2026-09-02. The v20.0.0 release is labelled "AI Staff positioning", which suggests the positioning change and the eight-bot team arrived together in that major version. If you pinned an earlier version, the jump to v20 is the one to read the notes for.
The README points to VERSIONS.md for version history and shows a generated version badge, so the release surface appears to be maintained deliberately rather than ad hoc. There is also a marketplace.json at the repository root and an openclaw.plugin.json, both of which are plugin manifests that will need to stay in step with releases.
Upgrade cost depends on how you installed it. A marketplace or skills-CLI install is a re-run of the same command at a new version. A Portable Lite install means downloading the new tar.gz asset and replacing the extracted plugin directory. A git clone means pulling and reconciling any local edits, which is the path most likely to produce merge friction if you have customized skills.
The licence is Apache-2.0. That permits commercial use, modification, and redistribution, and it includes a patent grant. It also requires that you keep the licence and notice files and state significant changes. This is a description of the licence text, not legal advice; if you plan to redistribute modified skills inside a product, have your own counsel review the notice requirements.
Editorial conclusion
Adopt Aaron Marketing Skills if your team already runs Claude Code, Codex, Cursor, or another supported host and wants marketing workflows as versioned Markdown rather than prompt snippets. Do not adopt it if you need a hosted dashboard, a single MCP server, or a guarantee that every bot tier works without your own data. Before installing, verify two things: that your host appears in docs/agent-compatibility.md, and that the Claude Code install path you choose is the marketplace one, since marketplace add alone registers the catalog without enabling skills.
Frequently asked questions
What is Aaron Marketing Skills?
It is a library of 120 marketing agent skills across seven disciplines (narrative, SEO/GEO, social, email, paid ads, influencer, and launch), distributed as a plugin, as portable skills on 70+ hosts, or as an eight-bot AI Staff on Grok Bot and Hermes Bot Mode. Skills and commands are plain Markdown, with small Bash and Python-stdlib runtimes for hooks, validation, scoring, and CI.
How do I install Aaron Marketing Skills in Claude Code?
The README gives a two-step marketplace flow: /plugin marketplace add aaron-he-zhu/aaron-marketing-skills, then /plugin install aaron-marketing@aaron. The README warns that marketplace add only registers the catalog, so the second command is what enables skills and commands.
Can I install Aaron Marketing Skills on hosts other than Claude Code?
Yes. The README lists Codex, Cursor, OpenCode, Antigravity, Gemini CLI, Copilot CLI, OpenClaw, Hermes, and 70+ other hosts, installed with npx skills add aaron-he-zhu/aaron-marketing-skills. A single skill can be installed with the -s flag, for example -s keyword-research.
What are the eight auditor gates in Aaron Marketing Skills?
They are TALE for narrative, CORE-EEAT and CITE for SEO/GEO, ECHO for social, SEND for email, ROAS for paid ads, STAR for influencer, and RAMP for launch. Each maps to a named auditor, such as content-quality-auditor or ad-account-auditor, and each framework is defined in the repository's references/ directory.
Does Aaron Marketing Skills add an MCP server?
No. The README states the plugin adds nothing to /mcp and that docs/mcp-catalog.json is a copy-paste reference only. The repository root is the authoring source of truth, not an Agent Plugins v1 install root.
Community notes