mstack

Marketing skills for AI coding agents.

mstack turns Claude Code or OpenAI Codex into a practical marketing operator: brand strategy, campaigns, lifecycle email, launch planning, SEO, content, ads, social, proof assets, reporting, and local marketing memory.

It is inspired by gstack, but the workflows are marketing-first and privacy-first.

What You Get

• 26 marketing skills.
• Claude Code support through ~/.claude/skills.
• OpenAI Codex support through ~/.codex/skills.
• Persistent per-project brand context under ~/.mstack/projects/<slug>/.
• No mstack telemetry, tracking, analytics, stable device IDs, or background service.

Install

Requirements:

• Git
• Claude Code, Codex CLI, or both
• Bun only if you want to regenerate skill docs from templates

Clone and run setup:

git clone https://github.com/VersoXBT/mstack.git ~/mstack
cd ~/mstack
./setup --host auto

Install for Claude only:

./setup --host claude

Install for Codex only:

./setup --host codex

Install for both:

./setup --host all

Setup writes local config to ~/.mstack/config.yaml, installs skill links into the selected host directories, and keeps telemetry: off plus update_check: false by default.

Skills

Foundation

Skill	Purpose
/m-brand	Brand workshop: voice, audience, positioning, competitors, channels
/m-competitive	Competitor analysis and market gaps
/m-positioning	Positioning and messaging framework
/m-icp	Ideal customer profile and buying triggers
/m-strategy	Marketing strategy and 90-day roadmap

Campaigns and Lifecycle

Skill	Purpose
/m-campaign	Integrated campaign plan, channel roles, asset map
/m-email	Lifecycle emails, nurture, onboarding, launch sequences
/m-launch	Product, feature, content, or campaign launch plan
/m-experiment	CRO, messaging, lifecycle, and creative test design
/m-proof	Testimonials, case studies, quote banks, proof blocks

SEO and Content

Skill	Purpose
/m-audit	Marketing audit across SEO, content, social, and brand
/m-keywords	Keyword research and content gaps
/m-brief	Content brief from topic or keyword
/m-write	Blog posts, articles, page copy, email copy
/m-edit	Brand voice and copy quality edit
/m-seo	On-page SEO optimization

Social, Ads, Utility

Skill	Purpose
/m-social	Platform-specific social posts
/m-threads	Twitter/X threads
/m-engage	Community engagement replies
/m-ads	Google, Meta, LinkedIn ad campaign structure and copy
/m-landing	Landing page copy
/m-repurpose	Turn one asset into platform-specific derivatives
/m-calendar	Content calendar planning
/m-report	Marketing performance report
/m-learn	Search local marketing learnings
/m-upgrade	Update mstack

Recommended Flow

1. Run /m-brand once per project.
2. Run /m-audit or /m-strategy to pick the highest-leverage work.
3. Use /m-campaign, /m-email, /m-launch, and /m-experiment to execute.
4. Use /m-report and /m-learn to turn outcomes into reusable memory.

Brand Context

mstack stores brand context per project:

name: "ArcTrace"
tagline: "Real-time on-chain analytics for builders"
voice:
  tone: "technical but approachable"
  personality: "builder-oriented, no-nonsense"
  avoid: ["corporate jargon", "buzzwords", "emojis"]
audience:
  primary:
    description: "DeFi developers and traders"
    pain_points: ["slow analytics", "fragmented data", "no real-time alerts"]
positioning:
  category: "on-chain analytics"
  differentiator: "real-time, builder-first"

Every skill can read this context, so copy, campaigns, reports, and strategy stay consistent across sessions.

Optional Data Sources

Skills work without API keys. When present, these credentials improve research and reporting:

Variable	Used for
SEMRUSH_API_KEY	Keyword and competitor research
AHREFS_API_KEY	Backlinks, keyword explorer, content gaps
GA4_CREDENTIALS	GA4 reporting
GA4_PROPERTY_ID	GA4 property selection
SEARCH_CONSOLE_CREDENTIALS	Search Console reporting
GSC_SITE_URL	Search Console site selection
OPENAI_API_KEY	Optional model-backed local workflows

Live browsing or API calls happen only when a workflow explicitly needs market research, competitor review, SERP checks, or reporting data.

Config

mstack-config list
mstack-config set skill_prefix true
mstack-config set proactive true
mstack-config set update_check false
mstack-config set telemetry off

telemetry: off is a compatibility key. mstack does not send telemetry.

Development

bun test
bun run gen:skill-docs --host claude
bun run gen:skill-docs --host codex
bun run skill:check

Generated Codex skills are placed under .agents/skills/ for local validation. Do not edit generated SKILL.md or agents/openai.yaml files by hand; edit the .tmpl source and regenerate.

Privacy

mstack does not collect telemetry, usage analytics, stable identifiers, or marketing content. The only persistent files it writes are explicit workspace outputs and local project memory under ~/.mstack/.

License

MIT.