- The Claude skills marketplace isn't an app store. It's an open catalog of versioned SKILL.md files you install with one command, and each one replaces a specific tool, ritual, or piece of manual work.
- 90 skills across 8 role-based packs plus Store Automation. Each is a Markdown file with inputs, framework, and output shape. Same input, same output, every run. Not a plugin, not a Claude feature toggle.
- Pick the skill that closes the biggest tab open on your desk today. Install with one npx command.
"Claude skills marketplace" is one of those phrases that sounds like an app store because we've been trained by app stores. It's not one. There's no paid tier, no featured shelf, no ratings inflation. The Claude skills marketplace is an open catalog of versioned SKILL.md files, installable into any Claude Code project with one command, and everything in it is Markdown you can read before you install it.
The useful way to browse the catalog isn't by name or by pack size. It's by what each skill replaces. If a skill doesn't obviate a tool subscription, a recurring ritual, or a chunk of manual work you already do, it's decoration. This post walks the 90 skills currently in the gtm-skills catalog, grouped into the 8 packs plus Store Automation, and names the specific thing each one is meant to replace.
The catalog is on GitHub, MIT-licensed, and free. The full skills directory with search, filters, and per-skill pages lives at /skills.
Why "what it replaces" is the right lens
Every one of these skills started as work someone was already doing. A pipeline audit, a cold email, an ad-account read, a margin calculation, a Monday KPI check. The point of turning it into a skill isn't automation - it's method. Two people running the same skill get the same output shape, and the skill you install today produces the same shape three months from now.
That's why the honest measure of a skill is what it takes off your calendar or off your Notion. If the-morning-ad-audit doesn't replace scrolling the ad account with coffee, it's a novelty. If the-lifecycle-mapper doesn't replace the broadcast list nobody trusts, it's a novelty. Read each row below as: "here's the thing I was already doing; here's the file that does it with a stated method instead."
The 8 packs plus Store Automation at a glance
| Pack | Skills | The desk this pack sits on |
|---|---|---|
| GTM Engineer | 7 | RevOps, marketing ops, systems |
| SDR | 9 | Outbound, lists, sequences, replies |
| Account Executive | 7 | Pipeline, deal execution, renewals |
| Experimentation Lead | 9 | CRO, test design, funnel diagnosis |
| Lifecycle Marketer | 7 | Segments, journeys, retention, save |
| Data Analyst | 10 | Weekly readouts, margin, cohorts, dashboards |
| Brand Designer | 5 | Voice, hooks, angles, scene direction |
| Performance Marketer | 27 | Paid social + paid search, angle to audit |
| Store Automation | 9 | Ecommerce loops, run on a cadence |
Performance Marketer is the biggest pack because paid media has more distinct daily jobs than any other GTM seat, not because it was padded. Brand Designer is the smallest because the pack sits deliberately upstream of decoration - voice, hooks, angles, scenes - not because there's less to do.
Product context - built inline, not a setup step
There's no separate setup skill any more. Every skill in the pack researches your business itself the first time you run one - it reads your site for positioning, offer, voice, and proof, asks at most three things research can't answer, and saves what it learned to .agents/product-context.md so the next skill doesn't repeat the work.
GTM Engineer pack - 7 skills
The plumbing between marketing and sales. Routing, workflows, transcript mining, enablement, competitor research. This is the pack that pays off when handoffs are broken and nobody can point at the rulebook.
| Skill | What it replaces |
|---|---|
the-routing-engine | A HubSpot or Salesforce workflow-builder session that never gets documented |
the-launch-gate | A Notion launch checklist nobody opens on launch day |
the-lead-router | LeanData rules maintained by one person who's on PTO |
the-workflow-builder | A Zapier canvas that breaks silently when a step fails |
the-transcript-miner | Gong summaries plus rep memory of what the prospect actually said |
the-enablement-kit | A Highspot library reps can't find the right template in |
the-competitor-dossier | A Klue tile that was accurate two product releases ago |
SDR pack - 9 skills
Outbound breaks in a predictable order: the list is wrong, so the opener can't be specific, so the sequence gets longer, so replies degrade, so nobody handles them. The pack attacks it in the same order it breaks.
| Skill | What it replaces |
|---|---|
the-list-builder | An Apollo search stapled to a Clay enrichment |
the-list-cleaner | A NeverBounce run and a dedupe formula |
the-fit-scorer | A HubSpot lead score nobody can defend |
the-monday-list | Scrolling Common Room, RB2B, or Warmly with no ranking |
the-cold-opener | A Lavender tab and a ChatGPT tab (subject line variants included) |
the-sequence-doctor | Outreach analytics you stopped reading after week two |
the-reply-classifier | Reading every reply by hand and forgetting which needed a fast follow-up |
the-call-booker | Four back-and-forth emails and a Calendly link |
the-no-show-save | The awkward "still on for today?" chase |
Account Executive pack - 7 skills
The part of sales that starts after someone agrees to talk. Pipeline reads, deal scoring (renewal risk included), objection prep, negotiation.
| Skill | What it replaces |
|---|---|
the-deal-gauge | A spreadsheet with Commit / Best Case / Worst Case columns (renewal risk in the same run) |
the-pipeline-scanner | Clari or Gong Forecast telling you what you already suspected |
the-call-coach | A manager's Friday afternoon of Gong clip review |
the-account-blueprint | A battlecard in Google Drive nobody rereads |
the-objection-playbook | A shared doc reps can't find mid-call |
the-win-loss-analyzer | A $30k outsourced win-loss study once a year |
the-negotiation-coach | The "we'll just see what they say" plan |
Experimentation Lead pack - 9 skills
CRO from "conversion is down" to a specific edit brief. Test design with real sample sizes, funnel leak diagnosis, page generation, personalization, pricing, activation, checkout.
| Skill | What it replaces |
|---|---|
the-hypothesis-engine | A whiteboard test-sizing session that ends in a shrug |
the-leak-finder | An Amplitude funnel report with no benchmark to anchor it |
the-page-shipper | A Figma handoff plus a two-week dev cycle for a launch page |
the-variant-router | Optimizely audience rules with no measurement plan |
the-price-point-finder | A Van Westendorp survey nobody's actioned |
the-first-mile-mapper | The "why aren't people activating" meeting that recurs quarterly |
the-spend-waste-finder | Two hours in Meta and Google reports looking for the leak |
the-checkout-auditor | A $9k Baymard audit for one flow |
the-pdp-reviewer | An unstructured "make this page better" ticket |
Lifecycle Marketer pack - 7 skills
Segments, journeys, retention, save. The pack breaks lifecycle into the jobs it actually is - segment, journey, campaign, save, referral, repeat - instead of pretending it's one job called "email."
| Skill | What it replaces |
|---|---|
the-lifecycle-mapper | A broadcast list nobody trusts to represent "active users" |
the-flow-architect | A Klaviyo canvas cobbled together over six edits |
the-campaign-engine | Three planning docs (copy, channel, cadence) that never align |
the-save-desk | A one-click cancel button and a hope-they-forget strategy |
the-referral-architect | A ReferralCandy install running on default settings |
the-repeat-purchase-check | The assumption that AOV will carry the second order |
the-promo-impact-check | "The sale did great" claim without the post-promo dip math |
Data Analyst pack - 10 skills
Raw exports into decisions. Margin stacks, cohort tables, weekly readouts, KPI blueprints, benchmarks, anomalies. Every skill has a stated method - the numbers don't come from a chart-eye squint.
| Skill | What it replaces |
|---|---|
the-kpi-blueprint | A Looker dashboard nobody defends the definitions on |
the-lever-finder | An OKR planning session with no maturity read behind it |
the-anomaly-alert | The "is this a real dip or just noise?" Slack thread |
the-benchmark-check | Googling "average X for SaaS" and quoting the first result |
the-cohort-tracker | A blended average that hides which cohort's actually broken |
the-margin-stack | A P&L that stops at gross margin and never shows CM1/CM2/CM3 |
the-shipping-margin-check | The assumption that free shipping is paying for itself |
the-inventory-risk-brief | The Sunday-night inventory reconcile |
the-returns-miner | A returns sheet with reason codes nobody groups by SKU |
the-weekly-readout | Five dashboards and no shared read on what actually changed |
Brand Designer pack - 5 skills
Voice, hooks, angles, scenes. The pack sits upstream of decoration on purpose - voice extracted from work you already like beats a fresh style guide nobody follows.
| Skill | What it replaces |
|---|---|
the-voice-fingerprint | A brand voice guide writers ignore in the moment |
the-angle-vault | A brainstorm that produces four rewordings of the same idea |
the-scene-composer | A photographer picking a location on gut feel |
the-hook-lab | Posting your first-draft opener and hoping the algorithm agrees |
the-activation-reel | A generic product tour video nobody watches to the end |
Performance Marketer pack - 27 skills
The newest and largest pack. It covers the whole paid loop - mining buyer language, writing angles, drafting campaigns as paused changes, auditing the account daily, diagnosing CPA moves across Meta and Google together. Every skill in the pack is read-only or drafts a paused change you approve by name. Nothing here spends money on its own.
| Skill | What it replaces |
|---|---|
the-brand-kit-reader | A discovery call to extract a brief that could have been read off the site |
the-ad-library-miner | Scrolling Meta Ad Library aimlessly for inspiration |
the-promise-sharpener | A headline that could belong to any brand in the category |
the-angle-spread | Six rewordings of the same idea labelled as five new angles |
the-copy-formula-picker | Copy written on "let's try a hook that feels good" |
the-creative-producer | A designer queue backing up while approved angles sit |
the-targeting-verdict | The reflex to stack more interests on a stalled ad set |
the-campaign-drafter | A launch that fragments budget across ad sets |
the-product-set-builder | Dynamic ads running on default catalog folders |
the-morning-ad-audit | Nobody looking at the account until Friday (fatigue call included) |
the-scale-pacer | Budget doubles that crash a winner in 24 hours |
the-angle-scoreboard | Scrolling results by ad instead of by the angle behind them |
the-pixel-audit | Trusting a conversion count that just moved without checking why |
the-search-term-miner | Letting Google spend on queries nobody actually read |
the-negative-keyword-builder | A broad -free negative that quietly kills real demand |
the-query-promoter | Copying every converting broad query to exact match on reflex |
the-search-intent-mapper | One ad group answering three different search jobs |
the-search-ad-writer | An RSA with headlines that go over the character limit |
the-quality-score-fixer | Raising the bid instead of fixing the ad |
the-conversion-goal-audit | Bidding on the wrong action for six weeks |
the-bid-strategy-picker | Switching strategies without auditing what you're bidding on |
the-search-scorecard | A weekly update with 30 numbers and no target on any of them |
the-search-week-review | A period-over-period comparison across unequal periods |
the-delivery-triage | Reaching for the budget slider on a stalled campaign |
the-change-plan-builder | A findings pile with no execution order |
the-cpa-diagnosis | Blaming one platform for a two-platform problem |
the-budget-reallocator | A platform split set by history, not by evidence |
Store Automation - 9 skills, not a 9th agent
Store Automation is deliberately not a 9th agent pack. The agent packs answer a question once; the loops run that answer on a cadence and diff it against the last run. That difference matters - a loop that quietly agrees with itself is worse than no loop at all.
| Skill | What it replaces |
|---|---|
the-loop-designer | A calendar reminder to "check the store on Monday" |
the-loop-ledger | Losing track of what a loop changed last week |
the-store-pulse | Morning reconciles of Shopify against two ad dashboards |
the-margin-sentry | ROAS looking stable while profit quietly drifts |
the-stockout-spend-guard | Paying to advertise SKUs you can't ship |
the-feed-watch | A Merchant Center disapproval you notice a week after sales dip |
the-catalog-audit | A sampled spot-check on a 10k SKU catalog |
the-launch-watch | Babysitting every launch's first two weeks by hand |
the-loop-auditor | A loop that quietly agrees with itself |
How to install the whole marketplace
One command installs the full catalog into your project's .claude/skills/ folder. Restart Claude Code afterward - skills load at startup, so a running session won't see them until the restart. Source and the per-skill SKILL.md files: sidchaudhary/gtm-skills.
$npx skills add sidchaudhary/gtm-skills -g --allThe -g flag installs at the user scope so every project you work in picks the skills up, and --all grabs every pack in one pass. If you only want one pack, drop --all and append the pack folder: npx skills add sidchaudhary/gtm-skills/skills/sdr. Pack folders: sdr, account-executive, gtm-engineer, experimentation-lead, lifecycle-marketer, data-analyst, brand-designer, performance-marketer, store-automation.
One skill worth trying if you're not sure where to start
If the pack-first framing feels like too much of a commitment, try one skill on real work you already have open. the-call-booker is a good candidate because the input is small (a warm reply on your desk), the output is immediate (a booking message, a reschedule, a confirmation, a day-before reminder), and the workflow it replaces is universal: the four-back-and-forth-emails-plus-a-Calendly-link dance that eats an hour every time a lead says yes.
Run it on the last warm reply you got. If the booking message it produces is worse than what you'd have written, that's a skill you can delete in ten seconds. If it's better - and for most SDRs it is, because the skill has read what "booking message that closes fast" actually looks like across a lot of them - keep it, and install the rest of the SDR pack next.
What the marketplace doesn't do
Skills don't execute. They produce strategy, definitions, drafts, and audits. When you need execution - to send the email, run the campaign, execute the SQL, actually move budget - the output has to plug into a system that runs. The Lifecycle pack's flows plug into Intempt Marketing. The SDR pack's cold openers plug into your existing outreach tool. The Data Analyst pack's margin stacks and cohort tables plug into Intempt Analytics or your own warehouse. Same principle across every pack: skills own the method, the run layer owns the run.
That's the reason the marketplace is worth treating as a marketplace at all. The skills are the reusable primitive between what you already do and what you already send. Install the pack that matches your desk, start with the skill whose "what it replaces" line hurts to read, and don't skip past the SKILL.md to the output.
Frequently asked questions. Answered.
The Claude skills marketplace is an open catalog of versioned SKILL.md files, published on GitHub and installable into any Claude Code project with one command. It's not a paid app store, it's not a plugin registry, and it's not a Claude feature you toggle on. A skill is a Markdown file with a stated method: inputs, framework, output shape. You install a pack of them into your project's `.claude/skills/` folder, restart Claude Code, and they show up as callable by name. The gtm-skills catalog covers 90 skills across 8 role-based packs plus Store Automation.
