Lenny's Product Skills Pack
86 product-management skills distilled from Lenny's Podcast, ready to install into Claude Code or any agent that reads SKILL.md files.
Evidence: The repository is a static collection of skill files with no code execution, so no permission requests, user confirmation, data flow, or sensitive data handling. External effects are limited to user-initiated file copying. Rollback mechanisms are not mentioned. Source attribution is clear: README and LICENSE state content originates from Lenny Rachitsky's podcast and newsletter, with IP retained. Deductions: No evidence for least privilege, user confirmation, data flow transparency, sensitive data handling, dependency security, external effects, or rollback, hence 0. Source attribution is explicitly stated but lacks verification mechanism, hence 2.
Evidence: The skill list in README is consistent with directory structure, no contradictions. Dependencies are only Markdown files, no external runtime dependencies, high availability. However, no failure messages or error handling are provided. Deductions: Self-consistency and dependency availability are supported, but failure messages are absent, hence 0.
Evidence: README clearly identifies target audience as product managers and AI agents, with broad scenario coverage. Capability boundaries are not explicitly defined, only implied by skill descriptions. Trigger precision is not defined, relying on agent's SKILL.md mechanism. Environment fit only mentions Claude Code, not other agents. Deductions: Audience and scenarios are well described, but capability boundaries, trigger precision, and environment fit lack evidence, hence 1.
Evidence: Information architecture is clear, organized by categories. Installation instructions are simple. Naming is stable and consistent. License is MIT, but content copyright belongs to Lenny. No examples or FAQ. Known limitations not listed. No versioning or changelog. Maintenance responsibility is with Refound AI, but update path is unclear. Deductions: Information architecture, install notes, naming stability, and license have evidence, but examples, known limitations, versioning, and maintenance responsibility lack evidence, hence 1.
Evidence: Output is Markdown skill files, directly usable by agents, high usability. Marginal value lies in providing verified quotes and frameworks, but no actual effectiveness data. Cost-benefit not mentioned, but as a free resource, cost is low. Deductions: Output usability and marginal value have evidence, but cost-benefit lacks specific data, hence 2.
Evidence: README claims all quotes are verbatim verified, but no verification method or tool is provided. Cross-source corroboration is not provided. Facts and inferences are not clearly separated. Deductions: Claim traceability has evidence, but cross-source corroboration and fact-inference separation are insufficient, hence 1.
- The repository is a static collection of skill files with no code execution, but users should review skill content themselves and avoid blind trust.
- The claim of verbatim quote verification lacks specific methodology; users should verify key quotes for accuracy.
- Skill files may contain outdated or inaccurate information; users should judge based on latest practices.
What does this agent do, and when should you use it?
This repository houses 76 product management and engineering skills (the README states 76, while the repo description says 86) distilled from the full archive of Lenny's Podcast and Newsletter, covering 597 episodes/posts and 4,019 sourced insights with verbatim quotes. Each skill is a standalone markdown folder containing a SKILL.md file (and possibly reference artifacts) that AI agents can load when working on a matching task. The skills are organized into a product operating model, with groups for strategy, planning, discovery, building, launch, growth, team, and operating cadence, plus vertical playbooks for marketplaces, AI products, zero-to-one, and enterprise/PLG. All quotes are verified verbatim, and each skill includes named frameworks, templates, and checklists. The project is MIT-licensed and aims to make Lenny's podcast wisdom actionable for AI agents.
It provides a collection of markdown skill files, each with a SKILL.md that gives AI agents specialized knowledge and workflows. Users clone the repo and copy skill folders into their .claude/skills/ directory, enabling Claude Code to load the relevant skill when a matching task appears. The skills cover a broad range of product management topics, each with frameworks, templates, and verified quotes from Lenny's podcast/newsletter (stored in references/artifacts.md). The agent reads these files and applies the frameworks during its reasoning. No code execution, APIs, or runtimes are included; it's purely a knowledge base for agents to consume.
- A product manager drafting a PRD can load the writing-prds skill to use proven frameworks and templates.
- A founder validating a new idea uses the idea-validation skill to run systematic checks before building.
- A growth lead applies the growth-model skill to build compounding loops beyond linear funnels.
- A team leader diagnosing a struggling team loads fixing-underperforming-teams to identify root causes and take action.
- A job seeker follows breaking-into-product to navigate the path into product management.
- A product marketer planning a launch uses launch-planning to turn a release into a remarkable event.
What are this agent's strengths and limitations?
- Backed by extensive research from Lenny's podcast/newsletter (597 episodes/posts, 4,019 sourced insights) with verbatim quotes, ensuring high credibility.
- Organized by product operating model, covering the full product lifecycle with vertical playbooks, making it comprehensive and easy to navigate.
- Every skill ships with named frameworks, templates, and checklists that are directly usable, not just abstract advice.
- MIT license permits free use and modification, while the underlying content is attributed to Lenny and guests, free and ungated.
- Primarily a text knowledge base – no code or pre-built automation, so agents must parse Markdown and apply content themselves.
- Installation requires manual folder copying without a package manager or auto-update mechanism, making updates tricky.
- Designed specifically for Claude Code; other agents may need adaptation to parse SKILL.md files.
- The skills are time-stamped to the podcast archive and may not include the very latest frameworks or trends.
How do you install or deploy this agent?
- Clone the repository: git clone https://github.com/RefoundAI/lenny-skills.git
- Copy a skill folder into your project's .claude/skills/ directory:
cp -R lenny-skills/skills/writing-prds .claude/skills/- Alternatively, grab individual skills via download links at https://refoundai.com/lenny-skills/.
No external dependencies are needed, but you must have an AI agent runtime that supports SKILL.md files (e.g., Claude Code).
How do you use this agent?
Once a skill folder is placed in .claude/skills/, Claude Code or any compatible agent will automatically load it when you work on a matching task. For example, if you add the writing-prds skill and then ask Claude Code to help write a PRD, it will reference the skill's frameworks, templates, and quotes. The skills are self-contained and require no extra configuration.