POPJAM Skills

Free, open-source agent skills for marketing, ads & creatives, shared by the POPJAM.IO team.

These are distilled versions of the pipelines we run in production at POPJAM: brand research, audience discovery, synthetic persona panels, ad creative generation, engagement simulation, landing pages that convert, and brand-guideline extraction. They work in Claude Code and any agent that supports the Agent Skills standard (Cursor, Codex, Amp, opencode, …).

License: MIT

Skills

Skill What it does
popjam-growth-engine Turn one website URL into tested ad creatives: brand research → audience segments → synthetic persona panel → ad angles → real creatives (image/video/animation) → deterministic engagement simulation → iteration. Ends with a leaderboard of who converts and why.
award-landing-page Design and build pages that sell (landing, pricing, waitlist, launch pages). Full deliverable: funnel research, art direction, conversion copywriting, distinctive design & motion, technical SEO, CTA wiring.
brand-guideline-extraction Reverse-engineer a complete brand style guide from a folder of design assets (posts, ads, banners, decks, videos), including a reusable "style prompt block" for on-brand AI image generation.
remotion-render-pitfalls Render-breaking Remotion mistakes that pass tsc + eslint but crash at render time. Battle-tested against POPJAM's production render pipeline; useful for any strict Remotion setup.

Install

With the skills CLI (recommended)

Works with Claude Code, Cursor, Codex, and every other supported agent:

# Install all skills (interactive picker)
npx skills add popjam-io/skills

# Or install a specific skill
npx skills add popjam-io/skills --skill popjam-growth-engine

Claude Code plugin marketplace

/plugin marketplace add popjam-io/skills
/plugin install popjam-skills@popjam

Installing the plugin also connects the POPJAM MCP server, so the hosted campaign tools show up in your session (browser sign-in on first use).

Manual (any agent)

Copy a skill directory into your agent's skills folder, e.g. for Claude Code:

git clone https://github.com/popjam-io/skills.git
cp -r skills/skills/popjam-growth-engine ~/.claude/skills/   # personal, all projects
# or into <your-project>/.claude/skills/ for a single project

Hosted MCP server

These skills run POPJAM's pipelines inside your own agent. If you'd rather call POPJAM's managed engine directly, connect the POPJAM MCP server: a remote server that exposes the same campaign tools (brand research, audience discovery, persona panels, ad generation, simulation) as callable tools, running on our infrastructure with durable memory.

Installing the plugin above already wires it up. To add it standalone in Claude Code:

claude mcp add --transport http popjam https://api.popjam.io/mcp

For Claude Desktop, claude.ai, Cursor, VS Code, and other clients, add https://api.popjam.io/mcp as a remote (streamable-HTTP) MCP server. Authentication is a one-time browser sign-in.

Full tool list and setup guides: popjam.io/mcp

Try it

Once installed, just describe the outcome and the skills trigger themselves:

  • "Make ads for https://yourbrand.com" → popjam-growth-engine runs the full pipeline
  • "Who is my target audience?" → audience segments + persona panel with evidence
  • "Test this ad concept against our audience" → persona simulation + scored leaderboard
  • "Build a landing page for our launch on Thursday" → award-landing-page
  • "Extract brand guidelines from ./assets" → brand-guideline-extraction

Some phases of the growth engine generate real media. Image/video generation uses the Higgsfield MCP and animations use Remotion. If neither is available, the skill still delivers strategy and simulation on text concepts, honestly.

Repo structure

skills/
├── popjam-growth-engine/
│   ├── SKILL.md              # entry point: frontmatter + instructions
│   ├── references/           # per-phase playbooks, loaded on demand
│   ├── scripts/              # deterministic scoring (never let the LLM do math)
│   ├── workflows/            # multi-agent fan-out templates
│   └── evals/                # eval suite the skill was validated against
├── award-landing-page/
├── brand-guideline-extraction/
└── remotion-render-pitfalls/

Each skill follows the Agent Skills specification: a SKILL.md with name + description frontmatter, and progressive disclosure, where references are read only when their phase begins, keeping context lean.

Why free?

We build POPJAM, an AI marketing platform that runs these same pipelines as a product (brand onboarding from a URL, reusable audiences & persona panels, ad/content generation, persona simulation with scored leaderboards, and campaign iteration), plus a team UI, asset storage, and integrations on top.

These skills are the "run it yourself in your agent" edition. If you'd rather have the whole loop managed for you (or your marketing team doesn't live in a terminal), try POPJAM.

Contributing

Issues and PRs welcome. See CONTRIBUTING.md. Good contributions: failure cases the skills should handle, new reference playbooks, eval cases, and new marketing/creative skills.

License

MIT © POPJAM