Skip to content

Latest commit

 

History

History

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 

README.md

igpt-marketing

Marketing intelligence for marketing leaders, brand managers, and growth operators. Turn the email already in your inbox into a live picture of agency deliverables, brand mentions, campaign feedback, event coordination, partnership pipeline, and press / PR activity.

6 Skills · 1 Agent · Structured Output · MIT License


Why this exists

A lot of marketing's signal lives in email — between the team, the agency, partners, journalists, customers, and the people who write inbound notes about your brand. Agency deliverables get committed to in threads that nobody centralizes. Customers say something quote-worthy in a reply and it never makes it into a case study. Press pitches outbound and the responses (or the silence) live in inboxes nobody audits at the end of the quarter. Event logistics scatter across dozens of vendors and confirmations. A grumpy customer email gets forwarded around but nobody flags whether it's a one-off or a pattern.

This plugin gives you six focused workflows that mine your marketing email and return structured answers about what's actually shipping, who's saying what about the brand, where partnerships and press are landing, and what's stalled.


What you can ask

Just ask in plain language. The agent or the right skill picks itself up based on what you say:

"Any agency deliverables outstanding?"

"What are people saying about us this quarter?"

"What's the press picture this month?"

"Where do partnerships stand?"

"Any event action items still open?"

"What feedback came back on the Spring launch?"

"Walk me through what needs my attention this week."

The last one — broad, cross-skill — is when the agent earns its keep. It runs the relevant skills in parallel and gives you a prioritized list of where to spend your attention.


A typical week

Monday morning agency check. "Any agency deliverables outstanding?" The agency-deliverables-tracker returns every deliverable from creative agencies and freelancers — what's in revision, what's late, what's awaiting feedback. Two from Atlas Studio are 4 days late; quick check-in.

Mid-week brand pulse. "What are people saying about us this quarter?" brand-mention-sentiment returns external mentions of the brand from customers, partners, and press, with sentiment patterns by audience type. You spot three quote-worthy customer praise lines worth sending to the case-study queue.

Pre-event sweep. "Any open action items for the Acme Summit?" event-action-items returns outstanding logistics, vendor deliverables, speaker confirmations, and post-event follow-ups for the upcoming event. Two speakers haven't confirmed travel and one sponsorship deliverable is overdue.

End of quarter, press + partnerships. "What's the press picture and where do partnerships stand?" press-coverage-tracker and partnership-pipeline-tracker give you the consolidated view for quarterly review.


The skills

Skill When to use it
agency-deliverables-tracker Deliverables expected from agencies and freelancers — revisions, feedback loops, what's late
brand-mention-sentiment External mentions of the brand by audience type; sentiment patterns, quotable lines
campaign-feedback-miner Feedback and reactions on specific campaigns from internal and external sources; common themes
event-action-items Outstanding tasks across event lifecycles — logistics, vendors, speakers, sponsors, post-event follow-up
partnership-pipeline-tracker Partnership and BD conversations — stage, last contact, next step, open commitments
press-coverage-tracker Press / PR activity — coverage landed, pitches outstanding, journalist relationships, media opportunities

Each skill collects what it needs from you (time window, agency, campaign, event, audience segment, etc.) and returns structured findings with evidence from the actual email threads. Quotes worth preserving are surfaced verbatim — voice is content.


The agent

Each plugin includes one agent that handles broad, cross-skill questions where you don't know which specific skill to invoke — "what needs my attention this week" or "how is the brand doing". The agent picks the right skills, runs them in parallel, and synthesizes a prioritized list — overdue agency deliverables and event blockers first, then press / brand signals worth responding to, then partnership pipeline items going stale.

The agent enforces two marketing-specific guardrails: quote brand mentions and customer feedback verbatim where it preserves voice (testimonial-worthy lines deserve to land in your hands as the customer wrote them), and distinguish signal from noise (a single grumpy email isn't a brand crisis; a pattern across multiple senders is).


What you see

These are real chat exchanges — you ask in plain language, Claude renders the answer as readable markdown with action items teed up. Quote-worthy customer voices are preserved verbatim. The structured output that powers the rendering is documented under "Output schemas" below if you want to see it.


You: any agency deliverables outstanding?

Claude: (running agency-deliverables-tracker — scanning agency and freelancer threads from the last 90 days)

5 deliverables in flight across 3 agencies. 2 are running late.

Late — Atlas Studio

Spring launch hero illustration (committed Mar 1, due Mar 5, 4 days overdue) Most recent thread: round 2 revisions sent Mar 3. Atlas confirmed receipt; no further update. Recommended: status check today; production deadline depends on this asset.

Customer-story video — Northstar Logistics edit (committed Feb 22, due Mar 8, 2 days overdue) Rough cut delivered on time; revised cut promised Mar 8 with the new B-roll; not yet received. Recommended: confirm new ETA; schedule the review session.

On track — Northwind Creative

Q2 campaign creative concepts (due Mar 14) Initial directions delivered Mar 6; feedback session scheduled for Mar 11. Status: on track.

Newsletter template refresh (due Mar 20) Wireframes approved Mar 4; design in progress. Status: on track.

On track — Beacon Design

Pitch deck template (due Mar 22) Outline approved Mar 7; design in progress. Status: on track.

Want me to draft the Atlas status check or the Northstar edit follow-up?


You: what are people saying about us this quarter?

Claude: (running brand-mention-sentiment — scanning inbound external email from the last 90 days for brand mentions and sentiment)

38 brand mentions surfaced across 4 audience types. Net positive overall; 3 quote-worthy lines and 1 pattern worth attention.

Customer praise — quote-worthy

"Honestly, this has saved my team about 4 hours a week. We didn't realize how much time we were spending on this until we stopped." — Operations Lead, Northstar Logistics, Feb 28

"The integration with our existing stack was the smoothest we've done in years. Worth flagging that to other prospects." — Head of IT, Apex Digital, Mar 3

"Your support team replied in 12 minutes on a Saturday. Genuinely didn't expect that and it mattered." — Customer success contact, Bluefield Co, Feb 14

Recommended: route these to the case-study and testimonial queue.

Press / analyst — neutral

Two analyst notes from independent firms referencing us in coverage of the category. Coverage is factual; no pull-quotes worth amplifying yet.

Customer concern — pattern worth attention

Three customers in the last 6 weeks raised the same friction point: the onboarding wizard's third step. Language varied ("confusing", "we got stuck", "took us a couple of attempts") but the consistency suggests a real UX issue rather than three unrelated complaints.

Recommended: brief Product about the pattern; one fix could prevent more friction.

Partner outreach — positive sentiment, no specific quotes

Three new partner inbounds in the last 30 days mentioning a peer recommendation. Suggests organic word-of-mouth in the partner ecosystem.

Want me to package the customer praise lines for Marketing's case-study queue, or draft a brief for Product on the onboarding wizard pattern?


You: what's the press picture this month?

Claude: (running press-coverage-tracker — scanning press / PR threads from the last 90 days)

Coverage landed — 3 stories. Pitches outstanding — 6.

Coverage that landed

Forbes — "How [Category] is changing" (published Mar 2) Featured quote from our CEO in a category overview piece. 8 companies mentioned; we got a single sentence. Modest amplification visible in the partner echo.

TechCrunch — funding announcement (published Feb 18) Standard funding coverage; Andrew picked up the story off the embargo list; quote from our CEO and from Beacon Partners as lead investor.

Stratechery — analytical piece referencing the category (published Feb 25) We're mentioned alongside three peers in a thesis-level piece; not a feature, but a category-credentializing mention.

Pitches outstanding

Outlet Reporter Pitch sent Status
The Information Saira Patel Mar 1 engaged — asked for customer references
Bloomberg Mark Chen Feb 22 acknowledged, no follow-up since
Wall Street Journal (no specific reporter) Feb 14 no response
Wired Lia Chen Mar 5 engaged — wants product walkthrough
Ars Technica Brian Diaz Feb 28 engaged — researching
Forbes Sarah K. Feb 8 gone quiet

Pitches needing action

  • The Information: provide the requested customer references this week to keep momentum.
  • Wired: schedule the product walkthrough.
  • Forbes (Sarah K.): close the loop or move on; 31 days quiet.

Want me to draft a customer reference brief for The Information, a Wired walkthrough scheduling note, or a close-the-loop check-in for Forbes?


You: walk me through what needs my attention this week

Claude: (the marketing agent kicks in, running 4 skills in parallel against your email through iGPT — agency deliverables, brand mentions, press coverage, event action items...)

This week, in priority order:

Production blockers (today)

  1. Atlas Studio — Spring launch hero illustration, 4 days late Production deadline depends on it. Status check today.

  2. Acme Summit — 2 speakers haven't confirmed travel Event is in 18 days. Travel locks in this week or next.

Press momentum (this week)

  1. The Information — customer references requested Saira Patel asked Mar 1; high-leverage piece. Provide references this week to keep momentum.

  2. Wired — product walkthrough scheduling Lia Chen engaged. Schedule before momentum cools.

Brand health (act this week)

  1. Onboarding wizard step 3 — pattern across 3 customers Brief Product. Consistent friction point worth a fix.

  2. Customer praise queue — 3 quote-worthy lines Route to case-study and testimonial queue.

Watch

  1. Forbes (Sarah K.) — 31 days quiet Close the loop or accept it's not happening.

Where would you like to start? I can draft the Atlas status check, the customer reference brief for The Information, the Wired scheduling note, or the Product brief on the onboarding pattern.


How it works

When you ask, the right skill picks itself (or for broad questions, the agent picks several). Each skill mines your connected email through iGPT — it searches relevant threads and asks structured questions about them. iGPT returns specific, evidence-backed findings; Claude then renders them into the readable answer you see in chat.

You ask in chat
        ↓
Agent (or skill) routes the question
        ↓
Skill queries iGPT against your connected email
        ↓
iGPT returns structured findings (with evidence)
        ↓
Claude renders the findings into readable markdown

You ask in plain language and the right answer comes back, grounded in your actual threads. The structured output that powers it is available in the schemas section below for anyone integrating directly.

Voice is content. When customer praise, campaign feedback, or brand mentions are quote-worthy, the renderer preserves them verbatim. A direct quote from the customer lands in your hands as the customer wrote it — useful for testimonials, case studies, pull-quotes, and internal proof points.

Patterns over outliers. A single grumpy email isn't a brand crisis. A pattern across multiple senders is. The skills and the agent flag patterns explicitly and avoid amplifying isolated reactions into something larger.

Your data stays with iGPT. The plugin doesn't pipe raw email to Claude — only the structured answers iGPT extracts. Your inbox is connected to iGPT through OAuth, not shared with the model.


Install

This plugin is part of the igpt-skills marketplace. Install it through your MCP client and connect your email — that's the whole setup.

Cowork

  1. Find the iGPT plugin marketplace. In Cowork, open the plugins section from the menu. You can either:

    • search the marketplace for iGPT, or
    • add it from GitHub: find the option to add a custom marketplace from a URL or GitHub repo and paste igptai/skills (or the full URL https://github.com/igptai/skills).

    Either path pulls the iGPT marketplace into Cowork so you can browse our plugins.

  2. Find igpt-marketing. In the iGPT marketplace, click into igpt-marketing to see the skills and the agent it includes.

  3. Install. Click install. Cowork sets up the plugin for you — no command line, no config editing.

  4. Connect your email through iGPT (one-time setup). The first time you ask the plugin a question, Cowork prompts you to sign in to iGPT and authorize email access. A browser window opens; you sign in to iGPT (or create a free account), pick which email account to connect (Gmail, Outlook, etc.), and approve the read-only permissions. iGPT does the email processing on its side so your inbox isn't shared with Claude — Claude only ever sees the structured findings iGPT returns.

  5. Ask anything. Once connected, just type questions in plain language — "any agency deliverables outstanding?", "what are people saying about us this quarter?" — and the right skill or the agent picks up automatically.

You do steps 1–4 once. After that, the plugin is always available and your iGPT connection is remembered.

Claude Desktop

Settings → ConnectorsAdd MCP Server → enter https://mcp.igpt.ai/ and complete the OAuth flow. The plugin's skills trigger automatically based on what you ask.

Claude Code

/plugin marketplace add igptai/skills
/plugin install igpt-marketing@igpt-skills

The plugin ships its own .mcp.json so the MCP server is registered for you. You'll be prompted for OAuth on first use.

Other MCP clients

Any Streamable-HTTP MCP client works. Point it at https://mcp.igpt.ai/ and use the prompts and JSON schemas from each SKILL.md directly.


Folder structure

igpt-marketing/
├── .claude-plugin/plugin.json
├── .mcp.json
├── README.md
├── agents/
│   └── igpt-marketing.md
└── skills/
    ├── agency-deliverables-tracker/
    ├── brand-mention-sentiment/
    ├── campaign-feedback-miner/
    ├── event-action-items/
    ├── partnership-pipeline-tracker/
    └── press-coverage-tracker/

Each skill folder contains a SKILL.md with the workflow (variables collected from the user, the iGPT query, output schema).


Customization

Time windows are natural language. Each skill asks how far back to scan. "Last 90 days", "since the rebrand", "May 2024", "since the campaign launched" — all fine. You don't need to type ISO dates.

Add categories your team already uses. Each skill's output schema has enums you can extend in the SKILL.md to fit your workflow — deliverable types specific to your agencies, audience segments specific to your ICP, campaign-feedback themes you track, partnership types, press tiers.

Add a new skill. Create a folder under skills/ with a SKILL.md that follows the project pattern (YAML frontmatter with name, description, metadata.version; a workflow that queries iGPT against your email; an output schema). Mirror the marketing guardrails — preserve voice in quotes, distinguish patterns from outliers. Add the new skill to the agent's "Available skills" list.

Use outside MCP. The prompts and schemas in each SKILL.md work directly through the iGPT API. The skills are documented at the file level, so you can lift their inputs and output schemas into your own code.


Output schemas (for developers / integrators)

The skills return strict, schema-validated output to the LLM, which is what produces the clean rendering you see above. Each schema is defined in its SKILL.md. If you're integrating directly with the iGPT API, here are two representative examples.

brand-mention-sentiment — for "what are people saying about us this quarter":

{
  "period_from": "2025-12-10",
  "period_to": "2026-03-10",
  "audience_scope": "all",
  "mentions": [
    {
      "audience_type": "customer",
      "person_company": "Operations Lead, Northstar Logistics",
      "date": "2026-02-28",
      "context": "Praising time savings",
      "sentiment": "positive",
      "quote": "Honestly, this has saved my team about 4 hours a week. We didn't realize how much time we were spending on this until we stopped.",
      "quote_worthy": true,
      "case_study_candidate": true
    },
    {
      "audience_type": "customer",
      "person_company": "Head of IT, Apex Digital",
      "date": "2026-03-03",
      "context": "Praising integration experience",
      "sentiment": "positive",
      "quote": "The integration with our existing stack was the smoothest we've done in years. Worth flagging that to other prospects.",
      "quote_worthy": true,
      "case_study_candidate": true
    }
  ],
  "patterns": [
    {
      "type": "consistent_friction",
      "summary": "3 customers in 6 weeks raised the onboarding wizard's third step",
      "audience_type": "customer",
      "evidence_count": 3,
      "severity": "moderate",
      "recommended_action": "Brief Product about the pattern"
    },
    {
      "type": "organic_word_of_mouth",
      "summary": "3 partner inbounds mentioned peer recommendations",
      "audience_type": "partner",
      "evidence_count": 3,
      "severity": "positive",
      "recommended_action": "Track for partner program signal"
    }
  ],
  "by_audience": {
    "customer": { "count": 18, "net_sentiment": "positive" },
    "press": { "count": 6, "net_sentiment": "neutral" },
    "partner": { "count": 9, "net_sentiment": "positive" },
    "prospect": { "count": 5, "net_sentiment": "mixed" }
  },
  "summary": "38 mentions, net positive overall. 3 quote-worthy lines for case studies; 1 pattern (onboarding step 3) worth a Product brief."
}

press-coverage-tracker — for "what's the press picture this month":

{
  "as_of": "2026-03-10",
  "coverage_landed": [
    {
      "outlet": "Forbes",
      "reporter": "Sarah K.",
      "headline": "How [Category] is changing",
      "published": "2026-03-02",
      "type": "category_overview",
      "depth": "single_sentence_mention"
    },
    {
      "outlet": "TechCrunch",
      "reporter": "Andrew M.",
      "headline": "Funding announcement",
      "published": "2026-02-18",
      "type": "funding",
      "depth": "feature"
    },
    {
      "outlet": "Stratechery",
      "headline": "Category thesis piece",
      "published": "2026-02-25",
      "type": "analytical",
      "depth": "category_credentializing_mention"
    }
  ],
  "pitches_outstanding": [
    {
      "outlet": "The Information",
      "reporter": "Saira Patel",
      "pitch_date": "2026-03-01",
      "status": "engaged",
      "outstanding_request": "customer references",
      "recommended_action": "Provide references this week to keep momentum"
    },
    {
      "outlet": "Wired",
      "reporter": "Lia Chen",
      "pitch_date": "2026-03-05",
      "status": "engaged",
      "outstanding_request": "product walkthrough",
      "recommended_action": "Schedule walkthrough before momentum cools"
    },
    {
      "outlet": "Forbes",
      "reporter": "Sarah K.",
      "pitch_date": "2026-02-08",
      "status": "gone_quiet",
      "days_quiet": 31,
      "recommended_action": "Close the loop or accept"
    }
  ],
  "coverage_count": 3,
  "pitches_outstanding_count": 6,
  "pitches_engaged_count": 3,
  "pitches_gone_quiet_count": 1,
  "summary": "3 stories landed. 6 pitches outstanding — 3 actively engaged, 1 gone quiet, 2 acknowledged without follow-up."
}

Related

License

MIT