# ai-marketing-skills: A Claude Code Skill Collection for Marketing and Sales Automation > ai-marketing-skills is an open-source repository of Claude Code skills covering growth experiments, content operations, SEO, outbound email, YouTube packaging, and finance automation. Each skill category ships its own scripts, configuration, and SKILL.md file. **ericosiu/ai-marketing-skills** — Open-source AI marketing skills — growth experiments, sales pipeline, content ops, outbound, SEO, and finance automation - Repository: https://github.com/ericosiu/ai-marketing-skills - Website: https://www.singlegrain.com - Stars: 3,596 · Forks: 689 - Language: Python - License: MIT - Published: 2026-09-23 · Updated: 2026-09-23 · Language: en - Canonical page: https://hysenlabs.com/projects/ericosiu-ai-marketing-skills ## What ai-marketing-skills Provides and Who It Is For ai-marketing-skills is described in the README as a collection of complete workflows rather than prompts. Each skill category contains scripts, scoring algorithms, configuration files, and a SKILL.md that tells Claude Code how to execute the workflow. The README describes the collection as built by the team at Single Brain and tested on real pipelines. The target user is a marketing or sales professional or engineer who uses Claude Code as an agentic coding assistant and wants to automate specific marketing tasks: running growth experiments, qualifying website visitors, scoring landing pages, generating video content derivatives, or producing financial models. The skills are not intended to be used without Claude Code; copying only the SKILL.md without the supporting scripts, references, and agents directories will leave the workflow incomplete, as the README explicitly warns. ## Repository Structure: Skill Categories and What Each Covers The repository organizes skills into over thirty directories, each representing a functional area. The README table describes them: Growth Engine covers autonomous marketing experiments that run, measure, and optimize themselves. Sales Pipeline handles turning anonymous visitors into qualified pipeline using tools like an RB2B Router, Deal Resurrector, and ICP Learner. Content Ops ships an Expert Panel, Quality Gate, and Editorial Brain for producing high-scoring content. SEO Ops includes Content Attack Briefs and a GSC Optimizer. Finance Ops runs cost estimation and scenario modeling. Video content has several dedicated categories: Video Content Engine for routing video into content portfolios; Shortform Production for producing clips via the API; YouTube Packaging for thumbnail and title packaging with performance tracking; and Agentic Video Understanding for goal-directed moment extraction from long video using Gemini. The Autoresearch category, described in the README as Karpathy-inspired optimization loops for conversion content, generates over fifty variants, scores them with an expert panel, and evolves winners. The X Long-Form and Humanizer skill includes a 24-pattern AI slop detector for posts written on X. Top-level files include a Makefile, a VERSION file, a CONTRIBUTING.md, and a skill-safety.yml. ## Getting Started: Clone, Pick a Category, and Configure The quick start pattern documented in the README applies to every skill category: ```bash git clone https://github.com/ericosiu/ai-marketing-skills.git cd ai-marketing-skills ``` Then navigate into a category and install its dependencies: ```bash cd growth-engine pip install -r requirements.txt ``` Copy and edit the environment file with API keys: ```bash cp .env.example .env ``` Then run the skill. The README shows a growth-engine example that runs the experiment engine: ```bash python experiment-engine.py create \ --hypothesis "Thread posts get 2x engagement vs single posts" \ --variable format \ --variants '["thread", "single"]' \ --metric impressions ``` For use with Claude Code, each skill directory must be copied in full into the Claude Code skill directory. The SKILL.md acts as the skill definition; Claude Code reads it when the skill is invoked. The README repeats the warning about copying only SKILL.md: supporting files in scripts/, references/, and agents/ subdirectories are required for the workflow to function. ## The Pre-Commit Security Hook The Makefile includes a setup target that installs a pre-commit security hook: ```bash make setup ``` This copies the script at security/pre-commit-hook.sh into .git/hooks/pre-commit and makes it executable. The hook runs before each commit to check for credentials or sensitive data before they reach version control. The presence of a security hook reflects a real concern for a marketing automation repository: skills that access sales pipelines, advertising APIs, and analytics services will handle API keys, access tokens, and potentially customer data. The skill-safety.yml file at the repository root documents safety guidelines. Engineers running these workflows in a team setting should review both files before connecting the skills to production data sources. ## Limitations: API Key Dependencies and Claude Code Requirement Nearly every skill category depends on external APIs. Video analysis skills reference Gemini. Sales pipeline skills reference services like RB2B. Growth and content skills may call advertising or analytics APIs. The .env.example files in each category document what keys are required, but the number of services involved means getting a full category running can take meaningful setup time. The skills are also explicitly designed for Claude Code. They are not standalone command-line tools that any developer can pick up without that context. A marketing team without a Claude Code setup will not be able to use these workflows as described. The README does not document which skills are ready for production use versus experimental. Some categories, such as Agentic Video Understanding and Net-New Video Editor, describe complex multi-step pipelines that the README notes can produce reversible, review-ready drafts. Teams should treat each skill category as requiring a validation pass against their own data before relying on it in a production pipeline. The VERSION file and per-skill SKILL.md files document which version each skill is at. ## ai-marketing-skills vs. Writing Individual Prompts The README's framing of these as skills rather than prompts reflects a real difference. A prompt is a single instruction; a skill bundles a prompt with scripts, reference data, expert panels, scoring algorithms, and multi-step orchestration logic. The Autoresearch category, for example, generates over fifty content variants, scores them using a configurable expert panel, and runs an evolution loop to improve winners. That workflow cannot be captured in a single prompt. The alternative approach, writing prompts on demand, requires the engineer or marketer to reconstruct the workflow logic each time. The trade-off is portability: ai-marketing-skills requires Claude Code, while a prompt-based approach works with any LLM interface. Teams already using Claude Code gain structured, repeatable workflows; teams without it should evaluate whether the setup investment is justified by the specific skills they need. ## Conclusion ai-marketing-skills is a practical starting point for marketing teams that have already adopted Claude Code and want structured workflows rather than ad-hoc prompts. It is less useful without a Claude Code setup, since the skills are designed to be copied into Claude Code skill directories rather than run standalone. Before using any skill in a production pipeline, read the SKILL.md, install the required dependencies, and set the documented environment variables. The repository requires API keys for several external services, which the quick start instructions ask you to configure via a .env file before running any workflow. ## FAQ ### What are the AI marketing skills in this repository used for? The skills automate specific marketing and sales tasks including growth experiment management, sales pipeline qualification, content scoring, SEO brief generation, YouTube video packaging, outbound email creation, and financial scenario modeling. ### Do I need Claude Code to use ai-marketing-skills? Yes. The skills are designed to be copied into Claude Code skill directories and invoked from Claude Code. The README warns that copying only the SKILL.md without supporting scripts, references, and agents directories will leave workflows incomplete. ### How do I install a skill from ai-marketing-skills? Clone the repository, navigate into the skill category directory, run pip install -r requirements.txt, copy .env.example to .env and fill in the required API keys, then copy the complete skill directory into your Claude Code skills folder. ## Sources - [ericosiu/ai-marketing-skills on GitHub](https://github.com/ericosiu/ai-marketing-skills) - [Issues](https://github.com/ericosiu/ai-marketing-skills/issues) - [License: MIT](https://github.com/ericosiu/ai-marketing-skills/blob/main/LICENSE) - [Project website](https://www.singlegrain.com) - [README](https://github.com/ericosiu/ai-marketing-skills/blob/main/README.md) --- Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/ericosiu-ai-marketing-skills