AgentStack
MCP verified MIT Self-run

Marketing Ai Stack

mcp-dapollonsky-marketing-ai-stack · by dapollonsky

A running list of open-source tools that try to do marketing work end-to-end. One pick per job, where I can find one.

No reviews yet
0 installs
12 views
0.0% view→install

Install

$ agentstack add mcp-dapollonsky-marketing-ai-stack

✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

Are you the author of Marketing Ai Stack? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

  ┏┳┓┏━┓┏━┓┏┓┏┓┏━┓╺┳╸╻┏┓╻┏━╸   ┏━┓╻   ┏━┓╺┳╸┏━┓┏━╸╻┏
  ┃┃┃┣━┫┣┳┛┣┻┓┣╸  ┃ ┃┃┗┫┃╺┓╺━╸┣━┫┃╺━╸┗━┓ ┃ ┣━┫┃  ┣┻┓
  ╹ ╹╹ ╹╹┗╸┗━┛┗━┛ ╹ ╹╹ ╹┗━┛   ╹ ╹╹   ┗━┛ ╹ ╹ ╹┗━╸╹ ╹

A running list of open-source tools that try to do marketing work end-to-end. One pick per job, where I can find one.

This list collects 18 open-source tools, MCP servers, and Claude skills grouped by the 8 workflows marketers tend to run: research, positioning, content, SEO, social, paid, lifecycle, and measurement. Each workflow breaks down into smaller jobs like keyword research, ad copy variants, or inbox triage. Where I found a pick I'm happy to recommend, it's listed as the canonical. 14 jobs don't have a canonical yet and are marked open.

How I try to pick. I look for tools that (1) fit a daily marketing workflow, (2) do the work itself rather than just suggest it, and (3) can be installed by a non-engineering marketer in under a week. When 2 tools serve meaningfully different use cases, I list both. Tools I considered and passed on go in [rejected/](rejected/) with my reasoning, so you can disagree out loud if I got one wrong. I try to re-review picks quarterly.

How to help. If a pick is wrong, a tool I missed should be here, or an open slot has been filled, open a PR against a YAML file in [data/entries/](data/entries/) and the README regenerates from there. Full flow in [CONTRIBUTING.md](CONTRIBUTING.md). Corrections welcome.

The shortlist

| Job | Install | Cancels | | ------------------------------ | ---------------------------------------------------------------------------------- | ------------------- | | Competitor monitoring | Open Scouts | Crayon | | Customer interviews | noScribe | Otter.ai | | Market research | GPT Researcher | Perplexity Pro | | Briefs, messaging, battlecards | Anthropic Marketing Plugin | Jasper Brand Voice | | Long blog posts | ALwrity | Jasper | | Keyword research | DataForSEO MCP | Semrush | | On-page SEO audits | SEO & GEO Claude Skills | Surfer SEO | | Site crawls | LibreCrawl | Screaming Frog | | Tracking LLM mentions | GEO AI Agent | Profound | | Social scheduling | Postiz | Buffer | | Long video to shorts | MoneyPrinterTurbo | Submagic | | Ad copy variants | claude-ads | Copy.ai | | Ad creative variants | ComfyUI | AdCreative.ai | | Ad budget management | Meta Ads MCP | Meta Ads Manager UI | | A/B test analysis | GrowthBook | Optimizely | | Inbox triage | Inbox Zero | Superhuman | | Analytics summaries | Google Analytics MCP | GA4 Explorations UI | | Performance reports | PyMC-Marketing | Nielsen MMM |

Open slots

14 marketing jobs where I haven't found an OSS pick I'm comfortable recommending yet. Closed tools tend to own these today (Clay for ICP discovery, Dreamdata for attribution, Profound for LLM citation tracking, Surfer for content updates, Modash for creator outreach). If you know of an OSS tool that fits, or you're building one, open a PR and I'll take a look.

ICP discovery · Customer personas · Updating old posts · Repurposing content · Case studies · Internal linking · Creator outreach · Community replies · Landing pages from a brief · Welcome email sequences · Email list segmentation · Newsletter writing · Deliverability monitoring · Attribution summaries.

Things I looked at and passed on live in [rejected/](rejected/) with my reasoning.


Why each pick won

🔭 Research & intelligence (canonicals + alternates)

Competitor monitoring

Canonical: Open Scouts. Firecrawl-backed agent loop that watches topics on a schedule and emails weekly briefs. A real inbox deliverable, not another dashboard to check. 🟢 Automate

Also considered:

  • changedetection.io. Deterministic page diffs via XPath and CSS selectors, with webhooks on every change. Best for cheap, reliable monitoring piped into a custom downstream agent. 🟢 Automate
  • FireGEO. Firecrawl starter for tracking brand and competitor citations inside ChatGPT, Claude, and Perplexity answers. Best when GEO visibility is the KPI, not web pages. 🟡 Augment

🗑️ [Rejected candidates](rejected/competitor-monitoring.md)

Customer interviews

Canonical: noScribe. Local-first interview transcription with speaker diarization and a usable review editor. No cloud hop for sensitive customer calls. Everything downstream (themes, quotes, pain points) is a single LLM pass. 🟢 Automate

Also considered:

  • Reddit Research MCP. MCP with semantic search across 20,000 subreddits and citations. Best for surfacing buyer language, objections, and phrasing from outside your own customer base. 🟢 Automate · 🔌 MCP

🗑️ [Rejected candidates](rejected/customer-voice-extraction.md)

ICP discovery

No canonical yet.

🗑️ [Rejected candidates](rejected/icp-discovery.md)

Market research

Canonical: GPT Researcher. Autonomous research agent that does tree-like topic exploration and returns a citation-backed brief. 26k stars, actively maintained. Describe the market area, get a structured report back. 🟢 Automate

Also considered:

  • LangChain Open Deep Research. Multi-agent research scaffold with MCP-native tool use and configurable report structures. For marketing engineers who want to tune the writer and planner stack. 🟡 Augment · 🔌 MCP

🗑️ [Rejected candidates](rejected/market-scans.md)

🧭 Positioning & strategy (canonicals + alternates)

Campaign briefs

Canonical: Anthropic Marketing Plugin. First-party plugin from Anthropic covering campaign briefs, launch messaging, and battlecards in 1 install. Ships with MCP wiring to Notion, HubSpot, Ahrefs, and Supermetrics. 🟢 Automate · 🔌 MCP

Also considered:

  • Corey Haines Marketing Skills. 36 Claude skills from a working SaaS marketer, structured around a shared product-marketing-context spine. Best for startup-flavored strategy rather than generic frameworks. 🟢 Automate

🗑️ [Rejected candidates](rejected/campaign-brief.md)

Launch messaging

Canonical: Covered by the Anthropic Marketing Plugin entry under Campaign briefs.

🗑️ [Rejected candidates](rejected/launch-messaging.md)

Battlecards

Canonical: Covered by the Anthropic Marketing Plugin entry under Campaign briefs.

🗑️ [Rejected candidates](rejected/battlecard.md)

Customer personas

No canonical yet.

🗑️ [Rejected candidates](rejected/icp-persona-docs.md)

✍️ Content production (canonicals + alternates)

Long blog posts

Canonical: ALwrity. Self-hosted blog drafting that ships the full pipeline: research, outline, SEO scoring, publish. 1 Streamlit app, 20-minute install. 🟡 Augment

Also considered:

  • Microsoft Content Generation Accelerator. Enterprise Azure AI Foundry pipeline with brief-to-asset agents and brand-compliance validation. For procurement-heavy orgs that need a Microsoft reference pattern. 🟡 Augment

🗑️ [Rejected candidates](rejected/long-form-drafting.md)

Updating old posts

No canonical yet.

🗑️ [Rejected candidates](rejected/refresh-decay-audit.md)

Repurposing content

No canonical yet.

🗑️ [Rejected candidates](rejected/repurposing.md)

Case studies

No canonical yet.

🗑️ [Rejected candidates](rejected/case-study.md)

🔍 Organic discovery (canonicals + alternates)

Keyword research

Canonical: DataForSEO MCP. Commercial-grade SEO data via MCP. Volume, CPC, difficulty, all agent-ready. The data foundation Ahrefs and Semrush sell, reachable from any MCP client. 🟢 Automate · 🔌 MCP

Also considered:

  • mcp-gsc. MCP for Google Search Console with query performance, URL inspection, and sitemap status. Best for diagnosing your own ranking traffic rather than prospecting new keywords. 🟢 Automate · 🔌 MCP

🗑️ [Rejected candidates](rejected/keyword-research.md)

On-page SEO audits

Canonical: SEO & GEO Claude Skills. Focused Claude skill pack with 20 SEO and GEO skills. On-page auditor runs against a URL from Claude Code with an EEAT and CITE scoring rubric. Narrow scope by design. 🟢 Automate

🗑️ [Rejected candidates](rejected/on-page-seo-audit.md)

Site crawls

Canonical: LibreCrawl. Screaming Frog replacement. MIT-licensed, unlimited URLs, full technical coverage for redirects, canonicals, schema, and Core Web Vitals. Runs as desktop or web app. 🟢 Automate

Also considered:

  • crawl4ai. LLM-friendly web crawler optimized for clean markdown and structured output. For marketing engineers composing custom SEO agents on top of a crawl layer. 🟡 Augment · 🔌 MCP
  • RustySEO. Rust and Tauri desktop SEO toolkit with a Nginx/Apache log-file parser LibreCrawl lacks. For log-based SEO forensics. 🟡 Augment

🗑️ [Rejected candidates](rejected/technical-seo-crawl.md)

Tracking LLM mentions

Canonical: GEO AI Agent. Tracks how a brand shows up in ChatGPT, Perplexity, and Gemini answers over time. Bright Data scraper under a CrewAI loop. Category is still unsettled, so expect this entry to move as better tools ship. 🟡 Augment

Also considered:

  • getcito. Self-hostable web app for GEO tracking with run-over-run dashboards. For teams that want a deployable tool rather than a Claude thread. 🟡 Augment

🗑️ [Rejected candidates](rejected/geo-aeo-citation-tracking.md)

Internal linking

No canonical yet.

🗑️ [Rejected candidates](rejected/internal-linking.md)

📱 Social & video (canonicals + alternates)

Social scheduling

Canonical: Postiz. Self-hosted social scheduler across X, LinkedIn, Instagram, TikTok, YouTube, Threads, and Bluesky. Content history and analytics stay on the team's own infrastructure. Settled atom. 🟢 Automate

Also considered:

  • LangChain Social Media Agent. LangGraph sourcing and drafting agent with Slack human-in-the-loop approval. For teams that want review queues before social sends rather than a calendar. 🟢 Automate
  • Mixpost. MIT-licensed social scheduler covering the same channels as Postiz. Best for commercial resale or closed-source embedding where AGPL would block you. 🟢 Automate

🗑️ [Rejected candidates](rejected/multi-channel-scheduling.md)

Long video to shorts

Canonical: MoneyPrinterTurbo. One-click AI short-video pipeline. Script, TTS, stock footage, subtitles, output. 55k stars, the category leader by a 100x margin. 🟢 Automate

Also considered:

  • OpenShorts. Clips existing long-form video into shorts with YouTube Studio integration and no watermarks. For podcast and YouTube repurposing rather than topic-from-scratch generation. 🟢 Automate
  • Viral Clips Crew. CrewAI template for transcription, highlight selection, and cutting agents. A reference pattern for custom video pipelines, not a shipped product. Author describes it as "vibe-coded," so expect to maintain any fork yourself. 🟡 Augment

🗑️ [Rejected candidates](rejected/long-to-shorts.md)

Creator outreach

No canonical yet.

🗑️ [Rejected candidates](rejected/creator-outreach.md)

Community replies

No canonical yet.

🗑️ [Rejected candidates](rejected/community-reply.md)

💰 Paid acquisition (canonicals + alternates)

Ad copy variants

Canonical: claude-ads. Claude skill pack for ad copy audit and variant generation. Covers Google, Meta, YouTube, LinkedIn, TikTok, Microsoft, and Apple with weighted scoring. 2.6k stars, narrow scope. 🟢 Automate

🗑️ [Rejected candidates](rejected/ad-copy-variants.md)

Ad creative variants

Canonical: ComfyUI. Node-based visual workflow app for generative image and video. Used by professional creative teams at OpenAI, Netflix, EA, and Amazon Studios for batch ad-variant generation. Settled atom. 🟡 Augment

🗑️ [Rejected candidates](rejected/ad-creative-variants.md)

Ad budget management

Canonical: Meta Ads MCP. Meta Business Partner-approved MCP for Facebook and Instagram ads. Clean API path and compliant credentials matter once an agent actually moves budget. 🟡 Augment · 🔌 MCP

Also considered:

  • Ads MCP (multi-platform). Single MCP covering Google, Meta, LinkedIn, and TikTok Ads through one endpoint. Tradeoff is lower per-platform fidelity. For agents that need unified multi-channel access. 🟡 Augment · 🔌 MCP
  • Google Ads MCP. Google's first-party Apache-2.0 MCP covering campaigns, budgets, keywords, ad groups, and performance queries. Covers Search and Performance Max alongside Meta Ads MCP. 🟡 Augment · 🔌 MCP

🗑️ [Rejected candidates](rejected/budget-rebalancing.md)

Landing pages from a brief

No canonical yet.

🗑️ [Rejected candidates](rejected/landing-page-from-brief.md)

A/B test analysis

Canonical: GrowthBook. Bayesian and frequentist experimentation platform with CUPED, sequential testing, SRM checks, and an official MCP. Handles LP and funnel tests end-to-end. Settled atom. 🟢 Automate · 🔌 MCP

🗑️ [Rejected candidates](rejected/ab-tes

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet — be the first.

Versions

  • v0.1.0 Imported from the upstream source.