Quick answer: the best starting stack for PMMs is free. Corey Haines’ marketingskills covers positioning context, launches, pricing and sales enablement; Anthropic’s official sales and marketing plugins add battlecards and win/loss reviews; and the small pmm-skillset repo handles ICP and message-market fit. Install commands for all 14 picks and a step-by-step PMM workflow are below.
Product marketing is mostly synthesis. You read forty call transcripts, six competitor pricing pages and a product spec, then turn all of it into one positioning doc, one battlecard and one launch plan. That is exactly the kind of work a Claude Code skill is built to repeat.
This guide covers the Claude Code skills for product marketers that are worth installing as of November 1, 2026. Every skill below was checked against its live source page: GitHub star counts and last-push dates come from the GitHub API, install commands come from each repo’s README, and paid prices come from the live checkout page. Where something could not be confirmed, it says so.
What are Claude Code skills for product marketers?
Short answer: A Claude Code skill is a folder with a SKILL.md file that tells Claude how to do one job, such as building a battlecard or running a positioning exercise. Skills live in ~/.claude/skills (all projects) or .claude/skills (one project), or arrive bundled inside plugins. Claude loads a skill only when your request matches its description.
For a product marketer, that means you stop re-pasting the same April Dunford prompt every quarter. The method lives in the skill. Your inputs (transcripts, competitor URLs, a spec) live in the project folder. Claude reads both.
If the concept is new to you, start with what Claude Code skills are and come back.
Why skills fit PMM work better than one-off prompts
| PMM task | Why a skill helps |
|---|---|
| Positioning | The framework stays fixed across quarters, so outputs are comparable |
| ICP and personas | The same four dimensions get filled every time, nothing skipped |
| Battlecards | A fixed template means sales sees the same layout for every competitor |
| Launch plans | Tiering, RACI and timeline sections are never forgotten |
| Customer research | Coding rules for transcripts stay consistent across batches |
| Win/loss | The same reason categories make quarter-over-quarter trends readable |
The common thread is consistency. A good PMM artifact is useful partly because it looks like the last one.
Three ways skills get onto your machine
npx skills add owner/repo: the open skills CLI. It copies skill folders into your skills directory.- Claude Code plugin marketplaces:
/plugin marketplace add owner/repo, then/plugin install name@marketplace. Only works in an interactive CLI session. - Manual copy:
git clonethe repo and copy folders into.claude/skills/.
Full walkthrough: how to install Claude Code skills.
Takeaway: a skill is a reusable method. For PMMs, the method is the asset.
The 14 best Claude Code skills for product marketers, by job
Short answer: For most PMMs, the best starting stack is Corey Haines’ free marketingskills (positioning context, launch, pricing, sales enablement, customer research), Anthropic’s official sales and marketing plugins (battlecards, win/loss), and the small pmm-skillset repo for ICP and message-market-fit work. Paid packs add breadth, not a different method.
Stats below were pulled from the GitHub API on 2026-11-01.
| Repo | Stars | Last push |
|---|---|---|
| ComposioHQ/awesome-claude-skills | 78,504 | 2026-09-18 |
| coreyhaines31/marketingskills | 57,803 | 2026-10-31 |
| anthropics/knowledge-work-plugins | 34,325 | 2026-11-01 |
| alirezarezvani/claude-skills | 29,110 | 2026-10-31 |
| deanpeters/Product-Manager-Skills | 7,823 | 2026-10-16 |
| OpenClaudia/openclaudia-skills | 724 | 2026-10-29 |
| Fearofsnakes/pmm-skillset | 30 | 2026-04-15 |
Stars measure attention, not quality. The pmm-skillset repo has 30 stars and is one of the most PMM-specific collections on GitHub.
Positioning and messaging
1. product-marketing (coreyhaines31/marketingskills). Free.
Source: github.com/coreyhaines31/marketingskills
The README calls this skill “the foundation”: it builds a product marketing context document, and every other skill in the repo checks it first for product, audience and positioning. Run it before anything else.
Install:
npx skills add coreyhaines31/marketingskills --skill product-marketingOr the whole set as a plugin:
/plugin marketplace add coreyhaines31/marketingskills
/plugin install marketing-skills2. positioning-statement and positioning-workshop (deanpeters/Product-Manager-Skills). Free.
Source: github.com/deanpeters/Product-Manager-Skills
Built for product managers, but the positioning workshop is pure PMM territory. The README lists 77 skills.
Install (from README):
/plugin marketplace add deanpeters/Product-Manager-Skills
/plugin install product-manager-skills@pm-skills3. marketing-strategy-pmm (alirezarezvani/claude-skills). Free.
Source: github.com/alirezarezvani/claude-skills (path marketing-skill/skills/marketing-strategy-pmm)
Covers positioning, go-to-market and competitive work in one skill. Directory listings describe it as using April Dunford positioning language (unconfirmed beyond those listings).
Install (from README):
/plugin marketplace add alirezarezvani/claude-skills
/plugin install marketing-skills@claude-code-skills4. message-market-fit (Fearofsnakes/pmm-skillset). Free.
Source: github.com/Fearofsnakes/pmm-skillset
Per the README, it audits messaging assets against one ICP persona, scores them across 5 pillars, generates 4 test variants and plans a 3-week test sprint. The repo has no plugin manifest, so install by copying:
git clone https://github.com/Fearofsnakes/pmm-skillset.git
cp -r pmm-skillset/skills/* .claude/skills/ICP and personas
5. icp-definition (pmm-skillset). Free. Builds a layered ICP across firmographic, behavioral, psychographic and language dimensions, plus an anti-ICP with disqualification signals. Same install as above.
6. proto-persona (deanpeters/Product-Manager-Skills). Free. A lighter persona builder for when you have hypotheses but little research. Same plugin install as #2.
Competitive intel and battlecards
7. competitive-intelligence and win-loss-review (anthropics/knowledge-work-plugins, sales plugin). Free, official.
Source: github.com/anthropics/knowledge-work-plugins
The README describes the sales plugin as able to “build competitive battlecards”. It also ships handle-objection, customer-voice and create-an-asset. The sibling marketing plugin adds competitive-brief and brand-review; the product-management plugin adds synthesize-research.
Install (from README):
claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install sales@knowledge-work-plugins
claude plugin install marketing@knowledge-work-plugins8. battle-card (pmm-skillset). Free. The README says it builds a one-page card from win/loss patterns, review data and pricing intel, with talk tracks, objection responses, landmine questions and a 90-second briefing, and that “every claim traces to evidence”.
9. competitor-profiling (marketingskills). Free. Researches competitors from their URLs. Pair it with the repo’s competitors skill, which builds comparison and alternative pages.
10. battle-card-builder (Product-Manager-Skills). Free. A second battlecard format if the pmm-skillset template does not match your sales team’s layout.
Launch plans
11. launch (marketingskills) and launch-brief (pmm-skillset). Free. The launch skill handles product launches and feature announcements. launch-brief goes further per its README: launch tiering, channel plan, success metrics, RACI, phased timeline and a 60-second pitch. Use one, not both, or you’ll get two competing plan formats.
Sales enablement, pricing and research
12. sales-enablement (marketingskills). Free. Pitch decks, one-pagers, objection handling docs and demo scripts, per the README.
13. pricing (marketingskills). Free. Pricing decisions, packaging and monetization. Alirezarezvani’s repo also has pricing-strategy and a pricing-strategist skill if you want a second opinion.
14. customer-research (marketingskills). Free. Conducts and synthesizes customer research. The README diagram shows it feeding copywriting, cro and competitors, which is how research ends up in messaging.
Takeaway: You can cover every core PMM job with free, actively maintained repos.
Paid Claude skill packs and vendor MCP servers
Short answer: Paid packs sell volume and curation. None of the three below is PMM-specific, and none publishes independent reviews we could find. Vendor MCP servers (Klue, Crayon) are different: they connect Claude to competitive data you already pay for.
| Product | Price (checked 2026-11-01) | What it is | PMM relevance |
|---|---|---|---|
| Inflectual Master Marketing Skills Bundle (Gumroad) | $167 one-time (page also offers 4 payments of $41.75) | Seller says 167 skills across 16 categories | Broad marketing; check the category list for PMM coverage |
| Ultimate Claude SEO & AEO Skills Pack (Contra, Samuel Uebersax) | $39 (shown against $79) | SEO and AEO auditing, planning, writing | Low for PMM; useful if you own comparison pages |
| SuperSkills | $50 one-time | Seller says 139 Claude Code skills | General; seller pages disagree on category count (15 vs 20) |
| Klue MCP server | Pricing not public | Lets Claude read (and per Klue, write back to) Klue CI data | High if you already use Klue |
| Crayon MCP server | Pricing not public | Connects Crayon competitive data to Claude | High if you already use Crayon |
Prices were read from the live pages’ price metadata. Feature counts are the sellers’ own claims and are unconfirmed.
We left out at least one heavily promoted bundle because its sales page uses countdown timers and “slots left” scarcity. If a skills pack needs a fake clock to sell, skip it.
When paying makes sense
- You want one folder with a consistent style instead of mixing five repos.
- You value someone else’s curation time more than $39 to $167.
- You already pay for Klue or Crayon and want battlecards built from live CI data instead of public web pages.
When it doesn’t
- You need PMM depth more than breadth. The free PMM repos above are more specific than any paid pack we found.
- You can’t read the skill files before buying. With free repos you can read every
SKILL.mdfirst.
Takeaway: buy a paid pack for breadth or curation, or an MCP server for data. Don’t buy one for a better positioning method.
Free vs paid: side-by-side comparison
Short answer: Free repos win on PMM specificity, transparency and update speed. Paid packs win on single-vendor consistency. Vendor MCP servers win on data access.
| Factor | Free GitHub repos | Paid packs | Vendor MCP servers |
|---|---|---|---|
| Cost | $0 | $39 to $167 one-time | Tied to platform contract, pricing not public |
| Read before you install | Yes, every file | Usually no | N/A (it’s a connector) |
| PMM-specific skills | Yes (pmm-skillset, marketingskills) | Not confirmed | Competitive intel only |
| Update cadence | Visible in commit history (most pushed in Oct 2026) | Seller-stated | Vendor-managed |
| Install | npx skills add or /plugin | Usually download and copy | Connector setup |
| Uses your own data | Only files you provide | Only files you provide | Yes, your CI platform data |
For a PMM, the sensible order is: install the free stack, use it on one real project, and only then decide whether a gap exists that money solves.
A starter stack by team size
| Situation | Install |
|---|---|
| Solo PMM at a startup | marketingskills + pmm-skillset |
| PMM supporting a sales team | Above + Anthropic sales plugin |
| PMM with Klue or Crayon | Above + that vendor’s MCP server |
| PMM who also owns comparison SEO | Above + competitors skill, see Claude Code skills for SEO |
How to run a PMM workflow with skills, step by step
Short answer: Put your raw inputs in one project folder, run the context skill first, then chain research, positioning, messaging and enablement skills, reviewing each output before it feeds the next.
The pmm-skillset README draws this chain explicitly: icp-definition feeds positioning-audit, which feeds voc-synthesis and competitor-teardown, which feed messaging-hierarchy. That order mirrors how most PMM teams already work.
Step 1: set up the project folder
pmm-q4/
.claude/skills/ # project skills
inputs/transcripts/ # interview and call transcripts (.txt or .md)
inputs/competitors/ # saved pricing pages, release notes
inputs/spec.md # the product or feature spec
outputs/Strip customer names and emails from transcripts before Claude reads them, unless your company’s AI policy allows otherwise.
Step 2: build the product context
Ask Claude to run the product-marketing skill. It writes a context doc the other marketingskills read first. Edit it by hand. Every downstream output inherits its mistakes.
Step 3: synthesize customer research
Point customer-research (or Anthropic’s synthesize-research) at inputs/transcripts/. Ask for themes with verbatim quotes and transcript filenames for each. Then spot-check five quotes against the source files. If a quote isn’t in the file, the synthesis is not trustworthy yet.
Step 4: positioning and ICP
Run icp-definition, then a positioning skill. Give it the research output, not a summary from memory.
Step 5: battlecards and win/loss
Run win-loss-review on closed-won and closed-lost notes, then battle-card per competitor. The pmm-skillset card expects evidence for each claim. Keep it that way: a battlecard with invented competitor weaknesses gets reps embarrassed on calls.
Step 6: launch and enablement
Feed the spec and the messaging doc to launch-brief, then sales-enablement for one-pagers and demo scripts. For the pricing page itself, pricing covers packaging logic and landing page copy skills cover the words.
Takeaway: the chain matters more than any single skill. Garbage context at step 2 flows into every artifact after it.
Use cases in detail: what each PMM job actually needs
Short answer: Each PMM job has one input that makes or breaks the output. Skills supply the method; you have to supply that input.
Positioning and messaging
The input that matters is real customer language. A positioning skill working only from your spec will produce the language in your spec. Feed it transcripts or review excerpts.
ICP and personas
The input is closed-won account data: industry, size, role, use case. Without it, icp-definition will produce a believable but invented ICP. Label that output “hypothesis” until data confirms it.
Competitive intel and battlecards
The input is dated evidence: pricing page snapshots, release notes, review-site excerpts, loss notes. Ask the skill to date every claim. Competitor pricing pages change, and a stale card is worse than none.
Launch plans
The input is a launch tier decision. Ask the skill to recommend a tier, but make the call yourself. A tier-1 plan for a tier-3 feature burns goodwill with sales and support.
Sales enablement
The input is the objection list from actual calls. Anthropic’s handle-objection skill and marketingskills’ sales-enablement both work better with ten real objections than with “common objections”. More in Claude Code skills for sales teams.
Pricing pages
The input is your packaging logic: what’s in each tier and why. Skills can draft tier names, comparison tables and FAQ copy. They cannot know your margins. Don’t let a skill pick price points.
Customer research synthesis
The input is the full transcripts, not notes. Claude Code reads local files, so long transcripts are workable, but very large batches may need splitting by segment. Ask for counts (“mentioned in 7 of 22 interviews”) so themes are weighted, then verify a sample.
Win/loss analysis
The input is a consistent reason taxonomy. Define 6 to 10 loss reasons once, write them into the skill or a reference file, and reuse them every quarter. That’s how trends become visible.
| Job | Must-have input | Common failure |
|---|---|---|
| Positioning | Customer verbatims | Echoes internal jargon |
| ICP | Closed-won data | Plausible fiction |
| Battlecards | Dated competitor evidence | Stale or invented weaknesses |
| Launch | Tier decision | Over-scoped plan |
| Enablement | Real objections | Generic rebuttals |
| Pricing page | Packaging logic | Skill guesses prices |
| Research | Full transcripts | Unverified quotes |
| Win/loss | Fixed reason taxonomy | Categories drift each quarter |
Common mistakes PMMs make with Claude Code skills
Short answer: The big ones are installing too many overlapping skills, skipping the context step, trusting synthesized quotes without checking, and shipping competitor claims without dates.
- Installing three launch skills. Overlapping skills compete for the same trigger, and you can’t predict which one loads. Pick one per job.
- Skipping context. The marketingskills README says every skill checks
product-marketingfirst. Without it, outputs are generic. - Not reading
SKILL.mdbefore installing. It’s a text file. Read it. It tells you what the skill will and won’t do. - Treating synthesis as evidence. A theme is a claim. The transcript is the evidence. Keep filenames attached.
- Undated competitive claims. Add “as of” dates to every competitor fact.
- Expecting
/pluginto work everywhere. The marketingskills README notes/pluginonly works in an interactive Claude Code CLI session. Usenpx skills addelsewhere. - Not customizing. The best PMM skill is often a fork of a free one with your template and taxonomy baked in. See how to build a Claude Code skill.
What this guide does not cover
- Non-Claude tools. Some repos above also ship Codex or Gemini variants, but install steps differ.
- Brand voice enforcement in depth: see Claude Code skills for brand voice.
- Wider marketing stacks: see Claude Code skills for marketers and skills for growth marketers.
Frequently asked questions
What are the best Claude Code skills for product marketing?
For free options, coreyhaines31/marketingskills (product-marketing, launch, pricing, sales-enablement, customer-research), Anthropic’s sales and marketing plugins, and Fearofsnakes/pmm-skillset cover nearly every PMM job.
Are Claude Code skills free?
Most are. Every GitHub repo in this guide is free. Paid packs exist, from $39 to $167 in our checks.
Can Claude Code create battlecards?
Yes, with a skill. Anthropic’s sales plugin README says it builds competitive battlecards, and pmm-skillset ships a battle-card skill.
Do I need to code to use Claude Code skills?
No. You run a few terminal commands to install, then talk to Claude in plain language. See Claude Code skills for non-developers.
What is the difference between a skill and a plugin?
A skill is one SKILL.md folder. A plugin is a package that can hold many skills plus commands and MCP connectors.
Where do Claude Code skills go?
~/.claude/skills for all projects, or .claude/skills inside one project.
How do I install the marketingskills repo?
npx skills add coreyhaines31/marketingskills, or /plugin marketplace add coreyhaines31/marketingskills then /plugin install marketing-skills.
Can Claude analyze customer interview transcripts?
Yes. Save transcripts as text files in your project and run a research synthesis skill. Verify quotes against the source files.
Can Claude do win/loss analysis?
Anthropic’s sales plugin includes a win-loss-review skill. Output quality depends on how consistent your loss notes are.
Is it safe to give Claude customer data?
That depends on your company’s AI policy and plan. Remove personal details from transcripts unless you are cleared to share them.
Are paid Claude skill packs worth it for PMMs?
Not for PMM depth, based on what we could verify. They may be worth it for breadth or if you prefer one curated source.
Does Klue or Crayon work with Claude?
Both have announced MCP servers that connect their competitive data to AI tools including Claude. Pricing is not public.
Can Claude Code write a pricing page?
It can draft structure, tier comparisons and FAQ copy with the pricing and copywriting skills. Price points should come from you.
How is a Claude skill different from a custom GPT?
A skill is a local file you can read, edit and version, and Claude loads it only when relevant. A custom GPT lives inside ChatGPT.
Can I build my own PMM skill?
Yes. A skill is a Markdown file with a name and description. Forking a free PMM skill and adding your template is the fastest route.
Do skills work in Claude Cowork too?
Anthropic’s knowledge-work-plugins README says the plugins are built for Cowork and compatible with Claude Code.
Sources and methodology
All checks were run on 2026-11-01.
- Repo stars and last-push dates: GitHub API (
gh api repos/OWNER/REPO). - Skill names and paths: each repo’s file tree via the GitHub API.
- Install commands: copied from each README (marketingskills, knowledge-work-plugins, claude-skills, Product-Manager-Skills, pmm-skillset, openclaudia-skills).
- Prices: live product pages and their embedded price metadata (Gumroad, Contra, SuperSkills site).
- Directories consulted: claudemarketplaces.com, Cult of Claude, ComposioHQ/awesome-claude-skills.
- Vendor MCP: Klue announcement, Crayon announcement.
Not done: hands-on output testing of each skill. Descriptions reflect README and seller claims.
Your next step
Install one free stack today: npx skills add coreyhaines31/marketingskills, plus the pmm-skillset folders. Run product-marketing on your real product, then one battlecard for your top competitor. Compare it to the card sales uses now. That one comparison tells you more than any list, including this one.
For the other roles on your team, see skills for founders and skills for marketing agencies.
