Image: GitHub
Why it mattersA marketer can now run the same AI agent a developer uses. One command points the agent at a business and the agent knows what to do next.
An agent inside Claude Code knows how to refactor a React component. It does not know how to write a landing page for a plumber in Phoenix. A new MIT-licensed skill pack tries to close that gap.
James Dickerson published boring-marketing-skills on GitHub on November 16. The repo had 545 stars and 210 forks two days later, measured on November 18 from the GitHub API. The pack carries 32 skills for AI coding agents (Claude Code, Codex and Cursor), grouped around a small-business marketing journey.
What is in the pack
The 32 skills split into seven groups.
- Start here, 5 skills: brand voice, positioning, audience and competitors.
- Get found, 8 skills: Google Business, local SEO, service pages, AI search visibility, site audits, keyword research, SEO content, programmatic SEO.
- Get chosen, 5 skills: reviews, mobile checks, landing pages, copy, lead magnets.
- Get the call, 6 skills: paid ads, Google tracking, ad testing, Meta, response speed, creative.
- Bring them back, 3 skills: email sequences, newsletter, performance.
- Win clients, 2 skills: finding local prospects, writing a pitch.
- Market yourself, 2 skills: social content, articles.
The pack is tool-agnostic. Every skill is written so it works with no paid software. Paid tools are optional add-ons.
The entry point is a slash command, /start-here. The user names the business. The skill reads the website and the reviews. It writes brand files that every other skill will reuse. Then it tells the user which skill to run next. The first run takes about three minutes.
Why it fits two scopes at once
The pack sits in two categories at the same time: new tools and generative-search work.
As a tool, it is a direct answer to the question of what a non-engineer uses a coding agent for. A marketer or a freelancer can type /start-here into Claude Code and have the agent pick the next action. The whole set was designed around this user, not around someone who already writes code for a living.
As a GEO and SEO artifact, the Get Found group names the surfaces that matter in 2026: local search, service-area pages, AI answer engines (ChatGPT, Perplexity, Google AI Overviews) and programmatic SEO. The Get the Call group has a skill that checks whether an ad campaign actually drove phone calls. The other option is that the ads took credit for calls that would have happened anyway.
What to watch
The repo is three days old. The real test is whether the output the skills produce reads like a real marketer wrote it. Dickerson's own site, boringmarketing.com, offers paid services as a different path. That is for buyers who want an agency rather than an agent. The disclosure is clear and the repo is MIT-licensed. The free path works end to end.
Nothing has been measured against a paid tool. The claim on the README that the skills come from "actual campaigns" is attributed to Dickerson. The fastest way to judge the pack is to run /start-here on a real business. Read the first skill's output as if a client had paid for it.
Source
boringmarketer/boring-marketing-skills, GitHub. Star count and the November 16 creation date were read from the GitHub API on November 18.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.
