skills
Health Warn
- License — License: MIT
- Description — Repository has a description
- Active repo — Last push 0 days ago
- Low visibility — Only 5 GitHub stars
Code Pass
- Code scan — Scanned 12 files during light audit, no dangerous patterns found
Permissions Pass
- Permissions — No dangerous permissions requested
No AI report is available for this listing yet.
Free agent skills for marketing, ads & creatives by POPJAM.IO — npx skills add popjam-io/skills
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: ad creative generation, target audience segmentation, reaction simulation with synthetic persona panels, campaign research, landing pages that convert, brand-guideline extraction, writing-style extraction, and ad-platform safe-zone validation. They work in Claude Code and any agent that supports the Agent Skills standard (Cursor, Codex, Amp, opencode, …).
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. |
| writing-style-extraction | Reverse-engineer a brand's verbal identity (voice, hooks, structure, CTAs, offer framing, emoji/hashtag policy, formality, language mix) from its existing captions, articles, ad copy and page copy, with a bundled stats script for the numbers and a per-channel "voice prompt block" for on-brand AI copywriting. |
| ad-platform-safe-zones | Current (2025–26, source-verified) safe-zone and canvas specs for ad creatives on Meta, TikTok, and Google placements, plus a bundled validator that checks finished images and renders overlay proofs. Catches the obsolete numbers everyone still uses — Meta's 9:16 bottom clear zone is now 35%, not 20%. |
| 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).
ChatGPT and Codex plugins
This repository is also a portable Agent Plugins package. The rootplugin.json and mcp.json bundle these skills with the hosted POPJAM MCP
server; .codex-plugin/plugin.json and .mcp.json remain as compatibility
fallbacks.
Codex and other portable-plugin hosts can load the package directly. Local
ChatGPT testing additionally requires registering the MCP endpoint in ChatGPT
developer mode and mapping the account-specific plugin_asdk_app... technical
ID in .app.json; that identifier is intentionally not committed. Public
directory availability follows OpenAI's review and publication flow. See
SUBMISSION.md for the submission-ready listing, review cases,
and final portal checklist.
Versioning
Plugin releases use semantic versioning. Keep the release version identical inplugin.json, .codex-plugin/plugin.json, the top-level version in.claude-plugin/marketplace.json, and that marketplace's popjam-skills
entry. Bump it on every published release and tag the commit as vX.Y.Z.
The 1.0.0 embedded in the Agent Plugins schema URLs and schema_version: 1
in chatgpt-app-submission.json describe file formats, not the POPJAM release.
Neither mcp.json nor .mcp.json has a plugin-version field.
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 for generating ad creatives, segmenting target audiences, and simulating reactions with synthetic personas, with supporting campaign research and durable workspace 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
- "Figure out our brand voice from these captions and make the ad copy sound like us" → writing-style-extraction
- "Will this creative get cropped or covered on Reels?" → ad-platform-safe-zones validates it and renders an overlay proof
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/
├── writing-style-extraction/
├── ad-platform-safe-zones/
└── 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
Reviews (0)
Sign in to leave a review.
Leave a reviewNo results found