agentleFS
PageproPagepro/nexity/.agents/skills/marketing-setup/SKILL.md

marketing-setup

Skill19 starsChanged 12 days agoMITView on GitHub

Walk a marketer with no terminal experience through setting up the marketing skills (brand-kit, site-contradictions, blog-post-quality-review, internal-links, blog-post-update, and source-review, which blog-post-update uses) step by step, by their preferences, whether they came as the Claude Code plugin `marketing-skills`, through `npx skills add Pagepro/nexity`, or in a downloaded ZIP of the repository. Checks where the skills are, says what to install and how (click by click), installs and builds what it can itself with permission, runs a short test of each chosen skill, and ends with what the person can now ask for. Use when someone says "Help me setup these skills for marketers", "set up the marketing skills", "skonfiguruj skille marketingowe", "pomóż mi to uruchomić", or has just installed the marketing skills and asks how to start.

  • Reads credentials
  • Installs packages

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When a step is fine, say so briefly and move on.

    line 33
  2. Distinctive:

    Ask these together, in one short message, and offer the defaults:

    line 37
  3. Very specific:

    Which website: should the tests use? Their own company site is best.

    line 50
  4. Very specific:

    For blog-post-update only: Search Console and GA4 data are optional.

    line 53
  5. Very specific:

    After an install, the agent app often does not see Node until it is restarted.

    line 107
  6. Very specific:

    Check node --version again before moving on.

    line 109
  7. Very specific:

    Only when blog-post-update is picked (it uses source-review to check facts), and <skills>/source-review/ts/dist/cli.js does not exist yet.

    line 119
  8. Distinctive:

    Run one test per picked skill, from the working folder.

    line 133

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Nexity Marketing Skills setup

The person in front of you is a marketer, not a developer. They installed the marketing skills (as a plugin, with npx skills, or by downloading the repository as a ZIP) and typed a sentence like "Help me setup these skills for marketers". Get them from there to working skills, one small step at a time.

You do the technical work yourself: you run the commands, read the output and fix what you can. The person only installs apps, clicks through dialogs and answers your questions.

Rules

  • Speak their language (the one they wrote in) and plain words. No jargon without a short explanation ("Node.js is a free program the skills run on").
  • One step at a time. Say what you are about to do and why in a sentence, do it, say what happened. When the person has to act (install an app, restart), give numbered clicks and wait for them to say "done".
  • Ask before you install or change anything on their computer. Running a read-only check (node --version) needs no permission; an install or a build does. Never use sudo, admin rights or a system package manager unless the person agrees and understands it.
  • Never ask for passwords, API keys or tokens in the chat, and never read or print .env files or key files. If a skill needs a key later, say where it goes, not what it is.
  • Do not edit the skills' files. The exceptions are the source-review build (step 3) and, in a ZIP, repairing .claude/skills (step 1). Working folders the skills create (.brand-kits/, .site-contradictions/, .source-review/, .internal-links/, .blog-updates/) are expected.
  • Stop at the first failure, explain it in one or two sentences and say exactly what the person should do, or what you will try next. Don't pile up errors.
  • When a step is fine, say so briefly and move on. Don't explain what already works.

Step 0: what they want

Ask these together, in one short message, and offer the defaults:

  1. Which skills? Describe each in one line of their language and let them pick (default: all):

    SkillIn plain wordsNeeds
    brand-kitA brand kit (voice, audiences, facts with sources) from a company's websitenothing extra
    blog-post-quality-reviewA critical review of one post or draft, with a 0–10 score and the three fixes that matter mostnothing extra
    site-contradictionsFinds every place where the site says two different things about the same fact (prices, numbers, plans, names), with a fix or a question for eachnothing extra
    internal-linksFinds links between the site's own articles that are missing, for the editor to addNode.js
    blog-post-updateWhy a post lost traffic and what to change in it, as a change list for the editorNode.js and a one-time build of source-review, which it uses to check facts; Chrome or Edge for images; Google data optional. Also uses blog-post-quality-review and work-humanizer
  2. Which website should the tests use? Their own company site is best. Ask for one blog post URL too, if they picked blog-post-quality-review or blog-post-update.
  3. For blog-post-update only: Search Console and GA4 data are optional. Don't set up Google access now; say that the skill asks about it on its first run and that it works without it (exports, or no data at all).

Step 1: where the skills are

<skills> below is the folder that holds this skill's folder: the parent of the folder this SKILL.md was loaded from. The other skills sit next to it (<skills>/source-review, …). Use its absolute path in every command. Work out how they were installed from that path, without asking:

  • Claude Code plugin (a path under ~/.claude/plugins/; the skills show as marketing-skills:<name>): nothing to check here. Tell the person that a plugin update replaces this folder, so source-review has to be built again after one (step 3).
  • npx skills (a path under .claude/skills/, .agents/skills/ or ~/.agents/skills/ outside a copy of the repository): check that every picked skill is there, plus work-humanizer when blog-post-update is picked. If one is missing, offer to add it with npx skills add Pagepro/nexity --skill <name> (ask first).
  • A copy of the repository (ZIP or clone; <skills> is <repo>/.agents/skills and <repo>/.agents/catalog.json exists):
    • The person should work in the root of that copy. If they opened a subfolder (base/, theme-meridian/, web/), the marketing skills are not there: tell them to open the folder one level up, the one made by unzipping (often nexity-main).
    • Claude Code finds skills in .claude/skills, which should be a link to .agents/skills. A ZIP unpacked on Windows (and by some unzip tools) turns that link into a small text file containing ../.agents/skills. If .claude/skills is a file, not a folder, ask to repair it, then replace the file with a copy of .agents/skills (on Windows, PowerShell: Remove-Item .claude\skills; Copy-Item .agents\skills .claude\skills -Recurse). Do the same for .claude/agents if it is a file. A repaired copy has to be repaired again after a newer ZIP (step 5). Then Claude Code needs a new session to see the skills: tell the person to start a new conversation in the same folder and write "continue the setup".
    • Codex reads .agents/skills directly; nothing to repair.

In any other agent, a skill is used by asking it to "follow <skills>/<name>/SKILL.md"; tell the person so.

The working folder is the folder the agent is open in. The skills put their results there. With a plugin or npx skills -g, suggest a folder of its own (e.g. Documents/Marketing AI) rather than the home folder or the Desktop, and tell them to always open that one.

Step 2: programs

Skip this step when they picked only brand-kit and blog-post-quality-review: those need nothing besides the agent, and the agent's web access (step 4 tests it).

Node.js 20.12 or newer (22 or newer for the images of blog-post-update). Check with node --version.

  • Missing or too old: the person installs the LTS version from nodejs.org (the big green button, then the installer with the default options). On a Mac it may ask for their computer password: that is the system's dialog, not you, and it is fine to type it there. If a package manager is already installed (brew on a Mac, winget on Windows), you may offer to install it for them instead (brew install node, winget install OpenJS.NodeJS.LTS); ask first.
  • After an install, the agent app often does not see Node until it is restarted. Ask them to quit the app completely, open it again in the same folder, and write "continue the setup". Check node --version again before moving on.
  • npm comes with Node; check npm --version when blog-post-update is picked (its source-review needs a build).

Chrome or Edge (only for blog-post-update, which renders images through the installed browser). Ask whether they have one; if not, the skill still works, only without images.

The skills do not need Git.

Step 3: build

Only when blog-post-update is picked (it uses source-review to check facts), and <skills>/source-review/ts/dist/cli.js does not exist yet. Ask, then run:

(cd "<skills>/source-review/ts" && npm install && npm run build)

It downloads its packages from the internet and takes a minute or two. Warnings from npm about deprecated packages or "vulnerabilities" are normal; only a line with ERR! or a failed build is a problem. A company network that blocks npm is the usual cause: say so, and ask them to try at home or ask IT to allow registry.npmjs.org.

Step 4: short tests

Run one test per picked skill, from the working folder. Each should take seconds; say what you are checking in plain words and report "works" or what failed.

SkillTestWorks when
anyList the skills you can see (Claude Code: type / and look, or your own skill list)Every picked skill is listed
brand-kit, site-contradictions, blog-post-quality-reviewOpen the test site's home page with your web toolYou can read its text (these skills need web access, not Node)
source-review (with blog-post-update)node "<skills>/source-review/ts/dist/cli.js" helpIt prints the command list
internal-linksnode "<skills>/internal-links/scripts/internal-links.mjs" check --site=<their site>It finds the sitemap and prints a sample article. Show them the first lines of the sample: if it is the menu or a cookie banner, that is fine for now; the skill fixes it on the first real run
blog-post-updatenode "<skills>/blog-post-update/scripts/blog-post-update.mjs" setupIt prints what is configured. "No Google access" is a valid answer

A failure here goes back to the step it belongs to (Node, the build, the folder). A site that blocks robots or renders with JavaScript only is a limit of the site, not of the setup: say so, and suggest another URL or pasting the text.

Then offer one real run of a cheap skill, so they see a result: a blog-post-quality-review of the post URL they gave, or a brand-kit of their site (longer, several minutes). Do it only if they say yes.

Step 5: what they can do now

End with a short summary in their language:

  1. What works and what was left out (and the one step to add it later).
  2. What to ask for, as sentences they can copy, one or two per skill they picked, e.g.:
  3. Where results land: in the working folder, in .brand-kits/, .site-contradictions/, .internal-links/, .blog-updates/ and .source-review/ (hidden folders: the agent opens the reports for them on request). Reports are HTML pages that open in the browser.
  4. Nothing is published. The skills read public pages and propose; a person makes the change in the CMS. internal-links can write Sanity drafts only when asked and given access.
  5. Updating, by how they installed:
    • plugin: /plugin → Marketplaces → nexity → Update (or turn on auto-update there), then "Help me setup these skills for marketers" once, to rebuild source-review;
    • npx skills: ask the agent to run npx skills update;
    • ZIP: download it again, unzip it next to the old folder, move the working folders above into the new one and say "Help me setup these skills for marketers" there.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.