Lenny Skills 2.0: A Markdown Skill Pack for Claude Code Product Work
86 product management skills from Lenny's Podcast for Claude Code and AI agents. Hiring, user research, strategy, shipping, and more.
At a glance
- What is it?
- The repository packages 76 product management skills drawn from Lenny's Podcast and Newsletter as SKILL.md folders you copy into .claude/skills/. The value is in the sourcing discipline and the artifact files, not in the agent integration, which is a plain file copy.
- Who is it for?
- Adopt it if you already run Claude Code on product work and want named frameworks with verbatim-sourced quotes sitting next to your code, and start by copying one skill such as skills/writing-prds into .claude/skills/ to see whether the agent picks it up on a real task. Skip it if you need a hosted tool, a database, or anything that updates itself, because this is a folder of markdown files you copy and re-copy by hand.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 61 days ago.
- What is it written in?
- GitHub does not report a main language for this repository.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
The Gap Between a Podcast Archive and an Agent's Context Window
Lenny's Podcast and Newsletter are large. The README puts the source base at 597 episodes and posts, with 4,019 sourced insights extracted from them. Nobody working on a pricing decision is going to re-listen to the relevant episodes before writing a doc, and an AI agent has no way to retrieve them at all unless someone converts the material into files the agent reads.
That conversion is the whole product. The repository's claim is narrow and checkable: each skill is a markdown folder that gives an agent specialized knowledge for a matching task, and every quote inside it was verified verbatim against the source transcript or post. The README is explicit that paraphrase drift is treated as a defect.
The audience is narrower than the topic list suggests. This is for people who already work inside Claude Code or another agent that reads SKILL.md files, and who want product frameworks to appear in the same context as the code and docs they are working on. If you do not use such an agent, the repository is a well-organized set of markdown notes with a manual copy step, which is a different and much less interesting thing.
What Actually Sits Inside a Skill Folder
The README describes a consistent layout per skill. There is a SKILL.md that the agent reads, and a references/artifacts.md that carries the named frameworks, checklists, and templates guests actually use. Version 2.0 added that artifacts file across skills; 1.0 did not have it.
The source counts printed next to each skill in the README tables are the most useful signal in the repository, because they tell you how much material stands behind a given folder. Roadmap Prioritization lists 43 sources. Defining Product Strategy lists 47. Product Taste and Intuition lists 36. At the thin end, Product Stack Strategy lists 6 and Product Experimentation Excellence lists 9. A skill built from 6 sources is a different artifact from one built from 47, and the tables make that visible without you opening anything.
2.0 also reorganized the set around a product operating model rather than a flat list: strategy, planning, discovery, building, launch, growth, team, and operating cadence, plus vertical playbooks and a career track. The README states that 1.0 was podcast-only and 2.0 folds in frameworks and templates from 349 newsletter posts, including the 'How X builds product' series. That is the substantive difference between the two releases, more than the skill count.
Installation Is a Copy Command, Not an Integration
There is no package, no CLI, and no installer. The README gives two commands:
git clone https://github.com/RefoundAI/lenny-skills.git cp -R lenny-skills/skills/writing-prds .claude/skills/
The destination is your project's .claude/skills/ directory. You copy the specific skill folders you want. The README also points to individual download links at refoundai.com/lenny-skills for grabbing single skills without cloning.
That is the entire setup. There is nothing to configure, no environment variable, no config key, no manifest to edit. Whether the agent then applies the right skill to a matching task depends on the agent's own skill-matching behavior, which the README asserts ('applies the right frameworks when you work on a matching task') but does not specify further. I cannot confirm from the material how selection works, how skills are ranked against each other, or what happens when two skills overlap. If you need deterministic routing, this repository does not describe it.
The practical consequence is that upgrades are manual too. When 2.0 arrived, anyone on 1.0 had to notice the release and re-copy folders. There is no mechanism described in the README for tracking which skills you copied or which version they came from.
The Skill Count Does Not Agree With Itself
The repository description says 86 product management skills. The README headline says 76, and the badge in the README also says 76. The v2.0.0 release is titled 'Lenny Skills DB 2.0' and is dated 2026-07-16, the same timestamp as the last push to main.
I cannot resolve that discrepancy from the supplied material, and it matters more than a typo would. If the description is stale, the README is authoritative and there are 76 skills. If the README tables are partial, the visible list understates what ships. Either way, the number you plan around should come from listing the skills/ directory after cloning, not from either text.
The same caution applies to the topic tags, which include pm-tools and product-skills alongside claude-code and llm. The tags describe the intended use, not a verified integration surface.
Where the Format Breaks Down
Skills are static text. A framework captured in references/artifacts.md in mid-2026 does not know about anything published afterward, and the README describes no update path beyond pulling a new release. For fast-moving areas this is the wrong tool. The AI-Assisted Prototyping skill lists 15 sources and Building With AI Agents lists 15; both cover ground where tooling changes faster than a curated markdown folder can track.
The second limitation is verification scope. The README's guarantee is about quotes: every quote checked verbatim against its source. That is a claim about fidelity to what a guest said, not about whether the advice works, generalizes, or applies to your market. A skill synthesized from 47 podcast appearances is still a synthesis of opinions, and the sourcing count tells you how many opinions, not how much evidence.
The third is the copy step itself. Because skills live in your project's .claude/skills/ directory, they are versioned with your code. A team that copies skills ad hoc will end up with different skill versions in different branches, and nothing in the repository prevents that. The README offers no guidance on this, which I read as an unaddressed gap rather than a deliberate choice.
Compared With Writing Your Own Skills
The obvious alternative is authoring your own SKILL.md files for the handful of workflows your team actually repeats. The difference is the work behind each folder. A hand-written skill encodes what your team already believes; a lenny-skills folder encodes what a set of named operators said across a large archive, with source counts attached so you can see the density.
That cuts both ways. Your own skill will match your terminology, your stage, and your constraints, and it will not carry someone else's assumptions about company size or go-to-market. Lenny Skills will be broader and better sourced, and it will include frameworks from companies whose context is not yours. The README's own framing supports this reading: the skills are grouped by process area so you can pull the ones that match, not adopt the whole set.
A middle path the repository supports directly: copy a skill, then edit references/artifacts.md to swap in your own templates while keeping the source quotes as reference material. Nothing in the layout prevents that, and it is the only way to get both the sourcing and your context.
Licence and Maintenance Cost
The licence is MIT, stated in the README badge and in the repository metadata. MIT permits commercial use, modification, and redistribution with the licence and copyright notice retained. I am not a lawyer and this is not legal advice; if you plan to redistribute modified skills inside a product, read the LICENSE file in the repository rather than the badge.
Maintenance cost is the cost of re-copying. There have been two releases: v1.0.0 on 2026-01-29 and v2.0.0 on 2026-07-16, roughly six months apart. If that cadence holds, expect a meaningful revision twice a year, each one requiring you to diff your copied folders against the new ones and re-apply any local edits to references/artifacts.md. That is the real recurring cost, and it grows with the number of skills you have customized.
The low-tech distribution is also what keeps the cost low. There is no runtime, no dependency tree, and nothing to break when the agent tooling changes, as long as the agent keeps reading SKILL.md files. If that convention shifts, the folders stop being skills and become plain documentation, which is a graceful failure rather than a hard one.
Editorial conclusion
Adopt it if you already run Claude Code on product work and want named frameworks with verbatim-sourced quotes sitting next to your code, and start by copying one skill such as skills/writing-prds into .claude/skills/ to see whether the agent picks it up on a real task. Skip it if you need a hosted tool, a database, or anything that updates itself, because this is a folder of markdown files you copy and re-copy by hand. Before committing, open references/artifacts.md in two or three skills and confirm the templates are specific enough to use, and check the 2.0 skill list against the 86 in the repository description, which does not match the 76 the README states.
Community notes