Marketing Skills for Claude Code: Copy, Pricing, Cold Email, Ads and Launch

Shipping a product is half the job. The other half is copy, pricing, outreach, ads and a launch plan, and a Claude Code session without marketing skills answers those from generic memory. The best-installed fix is copywriting by coreyhaines31/marketingskills, and the rest of the same repo covers the other jobs. (Search ranking is a separate topic, covered in the skills for marketing and SEO pillar.)

Key takeaways

• Copywriting leads with 239,051 installs (skills.sh registry), followed by marketing-psychology at 166,992 and ad-creative at 136,645 (skills.sh registry).
• Most of the strongest skills live in one repo, coreyhaines31/marketingskills, with 47,132 stars (GitHub), so one marketing context file serves all of them.
• Pricing has three options: pricing at 81,414 installs, phuryn/pm-skills pricing-strategy at 2,998, and refoundai/lenny-skills pricing-strategy at 2,253 (skills.sh registry).
• Outbound splits into two skills: prospecting (67,258 installs) builds the list and cold-email (129,826 installs) writes the sequence (skills.sh registry).
• For a conversion fix, cro has 92,775 installs (skills.sh registry) and ranks page problems by impact.

What is the best marketing skill for Claude Code?

Copywriting is the most installed, with 239,051 installs and 47,132 repo stars (skills.sh registry, GitHub). Install it with npx skills add https://github.com/coreyhaines31/marketingskills --skill copywriting.

Before drafting, the skill reads .agents/product-marketing.md if it exists, then asks for the page purpose, audience, offer and traffic source. It diagnoses the input as a blank brief, a draft with a real voice, or a point edit, and changes only what that situation calls for. Its own rules ban common AI tells in the copy it writes. Marketing-psychology (166,992 installs, skills.sh registry) pairs with it and works through mental models such as Jobs to Be Done and First Principles, each with a concrete marketing application. For voice cleanup after a draft, see the write like a human guide.

How do you get Claude Code to price a product?

Install pricing and let it ask about your market first. The pricing skill has 81,414 installs (skills.sh registry) and starts from value-based pricing: set the price between the next best alternative and the perceived value, and treat cost as a floor only. For a first price it suggests a rule of thumb of about $10, $100 or $1,000 per month depending on the customer, warns about the $9 trap, and points to Van Westendorp and MaxDiff for research.

Two smaller options suit different jobs. Pricing-strategy from phuryn/pm-skills (2,998 installs, 26,026 stars) compares flat-rate, per-seat, usage-based, tiered, freemium and value-based models, then designs two to four tiers. Pricing-strategy from refoundai/lenny-skills (2,253 installs, 1,306 stars) draws on interview quotes and recommends revisiting pricing every 6 to 12 months (skills.sh registry, GitHub).

How do you write cold emails and build a prospect list?

Use prospecting for the list, then cold-email for the message. Prospecting (67,258 installs, skills.sh registry) forks into four motions at intake: B2B SaaS, general B2B, local small businesses, and early demand-signal discovery. It insists on verifying emails before they reach a sequence and on licensed data instead of scraping LinkedIn.

Cold-email has 129,826 installs (skills.sh registry). It asks who you are writing to, what outcome you want, your proof and any research signals. Its rules: write like a peer, keep one ask, favor interest-based calls to action over meeting requests, and use short, boring, internal-looking subject lines. It also covers deliverability, such as secondary domains and turning off open tracking.

How do you run paid ads with Claude Code?

Pair a strategy skill with a creative skill. Ads (84,971 installs, skills.sh registry) builds platform-appropriate campaign plans from your goal, budget, landing page and audience. It routes to playbooks for Meta, LinkedIn and Google Search, and ends recommendations with facts, hypotheses and missing data, labeling unavailable inputs MISSING instead of guessing.

Ad-creative (136,645 installs, skills.sh registry) generates headlines and primary text by angle and checks character limits for Google, Meta and LinkedIn. It can also iterate from a pasted performance export. Before you publish anything, confirm the limits against the platform's own Google Ads help.

How do you plan a launch and fix conversion?

Launch gives you the plan and CRO diagnoses the page. Launch (80,179 installs, skills.sh registry) uses the ORB framework (owned, rented and borrowed channels), runs a readiness gate before any mechanics, and structures a release in five phases: prep, alpha, beta, early access and full launch. For a Product Hunt day specifically, product-hunt-launch from magentosh/skills (7,129 installs, 726 stars, skills.sh registry and GitHub) lists the specs: a 60 character tagline, up to 8 gallery images at 1270 x 760 pixels, and a maximum of 3 topics.

CRO (92,775 installs, skills.sh registry) scores a page on value proposition clarity first, then headline, call to action, scannability, trust signals, objections and friction. Output comes as quick wins and high-impact changes. To build the page itself, follow the landing page guide.

A complete starter stack

1. Copywriting: the base for every page and message (239,051 installs, skills.sh registry).
2. Marketing-psychology: framing and persuasion checks on the draft.
3. Pricing: set tiers and a first price before you launch.
4. Cold-email with prospecting: the list, then the sequence.
5. Ads with ad-creative: the plan, then the variations.
6. Launch and cro: the schedule, then the page fixes.

Add .agents/product-marketing.md with your audience, offer and proof once. Each skill reads it, so you answer the same questions one time. Skills load as described in the Claude Code skills docs.

Common pitfalls

• Skipping the context file, so every skill re-asks the same questions and invents answers when you rush.
• Sending cold email to an unverified list. Prospecting and cold-email both say to verify first, and a bad list hurts the sending domain.
• Treating generated ad variations as ready to run. Check platform limits and policies, and test one angle at a time.
• Launching before the readiness gate passes, which pushes attention at a product that is not ready for it.

Pick one skill for the job in front of you, run it on real product context, and add the next only when that job comes up.

Common questions