The marketing work you stop doing by hand.

For interactive marketing pieces: an ROI calculator or a data explorer built in React and bundled into a single self-contained HTML file.

Ten preset themes with hex codes and font pairings, shown as a PDF before you choose, then applied consistently across every slide or section.

The last link in a content chain: a finished draft reaches the CMS with no copy-paste, so headings and links survive the trip.

A CRM diagnostic that collects metrics across eight dimensions, grades each one, and produces a report you can hand to whoever owns the cleanup.

Word output with the formatting traps written down: table widths that break in Google Docs, shading that renders black, and a render-and-look verification step.

TEAM PICK

The connector under the analytics: 50+ tools across workflows, sheets and brand context, set up once and available in every Claude conversation after that.

TEAM PICK

Draft, test, evaluate, rewrite: the loop for turning one of your own repeated processes into a skill that behaves the same way every time.

TEAM PICK

The deck skill behind Claude's own PowerPoint output, including a validator that catches the chart and slide faults PowerPoint refuses to open.

TEAM PICK

The path from an API with no connector to a working MCP server, judged on whether an agent can complete tasks with it and not on endpoint coverage.

Campaign tracking settled at the source: a consistent naming scheme across the five UTM parameters, which is what makes attribution possible later.

Dashboard design for Looker Studio, separating the executive view from the operational one, with interactive controls and mobile layout treated as decisions.

A metric audit that fails a number for being a growing total, and demands a target with a pre-committed response for missing it.

Event tracking, UTMs and tag manager setup planned from the decision each event informs, with a naming scheme agreed before anything gets implemented.

TEAM PICK

The skill behind Claude's spreadsheet output: formulas and not hardcoded totals, a mandatory recalculation step, and a documented list of functions it refuses to write.

TEAM PICK

AI visibility data queried in plain English: share of voice, citation breakdown, sentiment by theme, and page-level performance in AI answers.

TEAM PICK

Search Console data answered in a sentence, including index coverage and URL inspection on the same connection.

TEAM PICK

Two tracks: choosing and reading an attribution model with no engineering, or building first-party attribution yourself when you own the page.

Referral and affiliate programs sized against your customer lifetime value and your current acquisition cost, so the incentive has a ceiling.

An activation audit that makes you define the aha moment as a single action, then removes every step standing between signup and reaching it.

Lifecycle email and SMS operations run through Klaviyo's own MCP server, so Claude reads your flows and revenue attribution over OAuth.

Churn work for a technical audience with a stated tone rule: no guilt trips and no desperate discounts, because developers read both as an insult.

TEAM PICK

Activation and retention diagnosed with the Fogg behavior model, where a missing prompt explains more churn than a lack of motivation ever does.

TEAM PICK

Sequence design across seven types, from welcome to win-back, with length and delay set by what triggered the person into it.

TEAM PICK

Cancel flows, save offers and failed-payment recovery treated as two problems: the customer who decided to go, and the card that quietly stopped working.

Registration flow work scoped tightly: how many screens, which fields are required, and which step loses people, before any copy gets touched.

Modals and overlays treated as a targeting problem: what the popup is for, what fires it, and how the existing one is currently performing.

A landing page audit built on three questions above the fold, with hero layout patterns matched to what the page is for.

Free tier design for products developers try before they buy, where a 14-day timer reads as a trap and a permanent limited tier does not.

Conversion work that starts with research and not with a hypothesis, mapping the funnel for high-traffic paths that convert badly and stages you never built.

TEAM PICK

A page audit ordered by impact, where value proposition clarity comes first and the 5-second test decides whether anything below it matters.

TEAM PICK

Experiment design with the math done up front: a sample size table showing that a 10% lift at a 1% baseline needs 150k visitors per variant.

Pitch decks, one-pagers, objection docs and demo scripts built from your differentiators, with the deal size and cycle length shaping what gets made.

The Salesforce outbound model as a scored diagnostic: role separation, a repeatable process, and a lead mix you can name before you plan pipeline.

An ICP framework that separates who fits from what makes them buy now, with must-have criteria written as filters a list builder can apply.

Enrichment plus judgment: it pulls firmographics, adds tech signals, scores against the ICP, and then explains each score in a paragraph a rep can read.

TEAM PICK

List building that asks who you sell to first, because a SaaS list, a local SMB list and a first-ten-customers list need different sources and different filters.

TEAM PICK

B2B cold email and follow-up sequences written to a peer standard, with a four-level personalization system and a test for whether yours is doing any work.

TEAM PICK

Lead scoring split in two so sales can sort by fit and marketing can trigger on engagement, built on HubSpot's current tool and not the retired score property.

Our paid stack, published as four skills against one Google Ads connector, with the negative keyword auditor the one that pays for the setup.

Pulls campaign, keyword and conversion data straight from the Google Ads API, so a spend question gets answered in the terminal instead of in a report builder.

Competitive intelligence from ad libraries that are already public, with frequency scoring on the pain points and a map of the positioning nobody has taken.

A two-step design skill: commit to an aesthetic position in writing, then express it on a canvas as a PNG or PDF you can put in an ad.

TEAM PICK

A spend audit that ends in an upload: exclusion lists grouped by theme, with statistical validation so you are not cutting on two clicks.

TEAM PICK

Campaign strategy with the target CPA and the budget asked for up front, plus conversion tracking, negative keywords and B2B account-based work.

TEAM PICK

Four modes from a single-batch generate to a monthly creative roadmap, with platform character limits enforced and a hard stop when the input folder is empty.

Two skills that run late in our content chain: one enforces brand voice across a draft, the other removes the recognizable AI writing patterns underneath it.

Outbound citations added in three phases, with a source tier ranking and one hard rule: no hallucinated links reach the draft.

Seven sequential sweeps over existing copy, each looking for one thing, with a loop back to check the earlier passes still hold.

A content pyramid that turns one pillar piece into 3 to 5 medium-form derivatives and then into short-form, with the effort marked at each tier.

A 25-node brief pipeline that scrapes the top 5 results, pulls organic keywords, and returns a writer-ready brief in under 10 minutes.

A voice document derived from your existing writing, with a tone matrix for how it moves between a crisis and a celebration.

TEAM PICK

A brief that tells a writer how long each section should be and which competitor gap it is filling, with a separate mode for pages that already exist.

TEAM PICK

A three-phase check that uses a different model per phase, because extracting claims, searching for sources and judging conflicting evidence are three different jobs.

TEAM PICK

Landing page and homepage copy written from a stated position: simple over complex, specific over vague, and a named list of buzzwords it will not use.

Structured data checked the way a search engine reads it, with a live list of which schema types Google still supports and which it has retired.

A refresh audit that treats listicles as a separate case, because the checks that improve a guide do very little for a ranked list.

Twelve programmatic page patterns with a matching rule for each, plus a defensibility ladder that tells you whether your data can carry a template at all.

Our own internal linking pass, documented with its thresholds, so you can rebuild the placement rules whether or not you use our stack.

Our AI-search audit, published with its shape: five subagents running at once, a composite score, and an action plan already sorted.

SEO for a technical audience, built around error-message queries and how-to-in-language searches, where thin content gets an instant bounce.

A gap analysis that ends in a sorted roadmap, because a list of keywords three competitors rank for is a research output and not a plan.

Turns AI-search citation data into a styled report, and separates prompts that cited you from prompts that pulled your page and named someone else.

TEAM PICK

An AI-search audit that scores passages for citability, checks whether AI crawlers can reach the page, and refuses the tactics Google has already dismissed.

TEAM PICK

Topic clustering built on SERP overlap, so two keywords group together because Google ranks the same pages for both and not because the words look similar.

TEAM PICK

One command crawls the site, spawns a specialist agent per area, and hands back a prioritized action plan tied to a health score.

TEAM PICK

Keyword research with the manual step removed: it pulls your current positions 4 to 15 as part of the same run, which is where the quick wins hide.

TEAM PICK

Answer engine work that starts with a baseline check across ChatGPT, Perplexity and AI Overviews before it recommends a single content change.

A platform matrix with the audience, the content style, the posting frequency and the one metric to watch for each channel you might use.

Reddit without the ban: a subreddit table with member counts and promo tolerance, plus the months of participation the channel expects before you post about yourself.

Earned media with the expectations set honestly: a press hit buys backlinks, legitimacy and AI-citation surface, and it will not buy you customers.

Docs written to rank, convert and retain: quickstarts cut time to value, references keep developers building, and good docs get shared where marketing pages do not.

Community strategy that starts from who members want to be, because a group organized around your product empties out once the novelty goes.

TEAM PICK

For anyone selling to developers: what works on Hacker News, what gets you flagged, and why the audience can tell when a post is marketing.

TEAM PICK

Content planning that separates search-driven pillars from shareable pieces, then maps keywords onto four buyer stages so the calendar has a reason behind it.

The SB7 framework as a scored diagnostic, with the customer cast as the hero and your brand as the guide who gets them through.

Pricing and packaging strategy with a Van Westendorp path for willingness to pay, plus a pricing page audit aimed at buyers and at the agents shortlisting tools.

Customer discovery scored against a rule that catches most research: state the progress a customer wants without naming your product or their demographics.

Offer construction before funnel work: raise the dream outcome and the perceived odds, cut the time delay and the effort, then score what you built.

TEAM PICK

The foundation skill: it creates .agents/product-marketing.md, and every other skill in this library reads it first so you stop repeating your positioning.

TEAM PICK

Positioning as a scored process: map the alternatives buyers weigh you against, then choose the market category that makes your strengths obvious.

TEAM PICK

An ICP build that ends with voice-of-customer language you can paste into copy, sourced from the communities your buyers already post in.