Best Claude Code skills for product managers in 2026

Last verified: 23 October 2026, against Anthropic's Claude Code documentation and the current contents of Anthropic's GitHub repositories.

Anthropic's skills documentation explains the idea in two sentences: "Skills extend what Claude can do. Create a SKILL.md file with instructions, and Claude adds it to its toolkit." The same page gives the reason a PM should care, which is cost: a skill's body loads only when it is used, so a long checklist or template costs almost nothing until you need it. Claude Code itself is not free; Claude's pricing page lists it from the Pro plan at $17 a month billed annually, or $20 billed monthly.

That makes the skill list the cheap part of the decision. The expensive part is the context every enabled plugin takes on every turn, so the list below is short on purpose.

How were these skills, plugins and subagents chosen?

Every pick had to meet four criteria:

• Anthropic publishes or documents it: in the Claude Code docs, in the official plugin marketplace, in the anthropics/skills repository, or in the knowledge-work-plugins repository.
• It does PM work, not only engineering work: specs, research, decks, reviews, stakeholder communication.
• It installs in one or two commands.
• Its contents were read on GitHub on 23 October 2026, not taken from a description.

Community skill collections were left out because they change weekly and nobody has reviewed them for you.

Which Claude Code skills and plugins should a PM install?

All of them are free to install; each needs Claude Code, which starts on Claude's Pro plan.

Skill, plugin or subagent	Best for	Free tier	Paid from	Standout
product-management plugin	Specs, research synthesis, roadmap and stakeholder updates	Free to install	Claude Pro, $17/month annual	Eight PM skills plus connectors for Linear, Jira, Notion, Figma, Amplitude
skill-creator	Building and testing your own skills	Free to install	Claude Pro, $17/month annual	Runs evals and tunes a skill's description
doc-coauthoring	Long documents: proposals, specs, decision docs	Free to install	Claude Pro, $17/month annual	Structured co-writing workflow that checks the doc works for readers
Document skills (docx, pptx, xlsx, pdf)	Decks, spreadsheets and Word files for stakeholders	Free to install	Claude Pro, $17/month annual	The same skills behind Claude's file creation
internal-comms	Status reports, leadership updates, FAQs	Free to install	Claude Pro, $17/month annual	Uses your company's preferred formats
playground plugin	Interactive explainers and quick concept demos	Free to install	Claude Pro, $17/month annual	Single-file HTML explorers with live preview
feature-dev plugin	Scoping a feature against a real codebase	Free to install	Claude Pro, $17/month annual	Explorer, architect and reviewer subagents
Built-in Explore and Plan subagents	Codebase research without flooding your session	Built in	Claude Pro, $17/month annual	Read-only, separate context window

Why is the product-management plugin the best place to start?

Anthropic's product-management plugin is the closest thing to a ready-made PM toolkit for Claude Code. Its README on GitHub describes it as covering "the full PM workflow: writing feature specs, managing roadmaps, communicating with stakeholders, synthesizing user research, analyzing competitors, and tracking product metrics." The plugin folder holds eight skills: write-spec, synthesize-research, competitive-brief, metrics-review, roadmap-update, stakeholder-update, sprint-planning and product-brainstorming. Its connector file points at hosted MCP servers for Slack, Linear, Asana, Monday, ClickUp, Atlassian, Notion, Figma, Amplitude and others.

Install it from your shell with claude plugin marketplace add anthropics/knowledge-work-plugins, then claude plugin install product-management@knowledge-work-plugins. Plugin skills appear as /<plugin>:<skill>, so the spec skill runs as /product-management:write-spec. Treat the output as a first draft written in Anthropic's template; the repository README says the plugins become much more useful once you customize them for how your company works.

Is skill-creator worth installing if you only use other people's skills?

Yes, because the first skill you will want is one nobody has written: your company's PRD template, your launch checklist, your interview synthesis format. skill-creator is published both in the anthropics/skills repository and as a plugin in Anthropic's official marketplace. Its SKILL.md description says it creates new skills, improves existing ones, runs evals and optimizes a skill's description "for better triggering accuracy", which is the part most hand-written skills get wrong.

The Claude Code skills for product managers guide walks through writing a PM skill by hand, which is worth doing once before you automate it.

Which skills help a PM write documents and decks?

Three skills from the anthropics/skills repository cover most PM writing. doc-coauthoring guides a structured workflow for documentation, proposals, technical specs and decision docs, including a step that checks the document works for its readers. internal-comms covers status reports, leadership updates, company newsletters, FAQs, incident reports and project updates. The docx, pptx, xlsx and pdf skills are the ones the repository README says power Claude's document capabilities; they are source-available rather than open source.

To add them in Claude Code, register the repository as a marketplace with /plugin marketplace add anthropics/skills, then install document-skills or example-skills from it, as the repository README describes.

What do subagents add for product work?

Subagents keep research out of your main conversation. The subagent documentation says each one "runs in its own context window with a custom system prompt, specific tool access, and independent permissions", and returns only a summary. Claude Code ships with built-in subagents: Explore, a fast read-only agent for searching a codebase, Plan, which researches during plan mode, and a general-purpose agent for multi-step work.

For a PM, Explore is the useful default: ask how a feature works today and Claude searches the code without filling your session with file contents. Write a custom subagent only when you keep spawning the same kind of worker with the same instructions; the Claude Code subagents for PM workflows guide has examples.

When is the feature-dev plugin useful for a PM?

feature-dev is for a PM who works inside the repository and wants a feature scoped before any code is written. Its README in the official marketplace describes a 7-phase workflow started with /feature-dev, and the plugin ships three subagents: code-explorer, code-architect and code-reviewer. The early phases ask clarifying questions and map the codebase, which is where a PM's requirements either survive contact with the code or do not.

The playground plugin is the lighter alternative when you need to show an idea rather than build it: its skill creates self-contained, single-file HTML explorers with visual controls and a live preview.

Which Claude Code skills are best for a solo PM or founder?

Install the product-management plugin and skill-creator, and stop there for the first two weeks. A solo builder pays for every enabled plugin in context on every turn, per Anthropic's plugin documentation, so add the document skills only when you produce decks or spreadsheets weekly. The bundled /doctor skill flags unused skills, MCP servers and plugins against their context cost, which makes it a good monthly cleanup.

Which Claude Code skills are best for a product team?

Install at project scope so the whole team gets the same setup. Anthropic's install documentation says project scope writes the plugin to the repository's .claude/settings.json, which you commit, and each collaborator then runs the install command once. A team gets the most from one shared, customized product-management plugin with your own PRD template and connectors, rather than each PM installing a different set.

What is the risk of installing community skills?

A plugin is code you run. Anthropic's official marketplace README warns: "Make sure you trust a plugin before installing, updating, or using it." It adds that Anthropic does not control what MCP servers, files or other software are included in plugins. Read a third-party skill's SKILL.md and any hooks before you install it, and prefer project scope so a teammate can review what changed. The Claude Code permissions and safety guide covers what to restrict.

Where can you learn to build your own Claude Code system?

Builders Camp's Building with Claude Code bootcamp teaches the layer underneath these plugins: moving from prompting to production-ready AI systems built with context, skills, agents and automation pipelines. PMs newer to the tool can start with Claude Code for Product Managers, and the CLAUDE.md vs skills vs hooks vs subagents guide explains where each rule belongs before you install anything.