lean-ux
Apply lean thinking to UX: hypothesis-driven design, collaborative sketching, and rapid experiments instead of heavy deliverables
Install / Use
npx skills add wondelai/skills --skill lean-uxInstalls into whichever agent you are using.
SKILL.md
Installable skill definition
Quality Score
Category
Content & MediaSupported Platforms
Our assessment of lean-ux
lean-ux scores 87/100 on our quality scale, 644th of 1,155 Content & Media skills we index.
Its SKILL.md is 16 KB long, well organised into 14 sections and no code examples: a thorough specification that gives an agent plenty to work with.
With 2,248 GitHub stars, it is one of the more widely adopted skills in the catalogue.
Maintenance, license and trust
- The repository was last updated about 2 months ago, so lean-ux is actively maintained.
- Our last check on 2026-11-02 found the source still online.
- It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
- Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-09-30. Automated pattern scan on 2026-09-29. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
lean-ux compared with similar skills
All 4 of these similar skills score higher than lean-ux; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| lean-ux (this skill)by wondelai | 87 | 2.2k | 2mo ago | SKILL.md |
| taste-skillby Leonxlnx | 100 | 116.8k | 8d ago | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 112.3k | 8d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 79.8k | today | CLAUDE.md |
| Claude-Code-Game-Studiosby Donchitos | 100 | 39.7k | 7d ago | CLAUDE.md |
Frequently asked questions
- How do I install lean-ux?
- Run
npx skills add wondelai/skills --skill lean-ux. The install tabs above show the steps for each supported agent. - Which AI agents does lean-ux work with?
- It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
- Is lean-ux safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is lean-ux still maintained?
- The repository was last updated about 2 months ago, so lean-ux is actively maintained.
Skill content
View source on GitHubname: lean-ux description: 'Apply lean thinking to UX: hypothesis-driven design, collaborative sketching, and rapid experiments instead of heavy deliverables. Use when the user mentions "Lean UX", "design hypothesis", "outcome over output", "design studio method", "assumption mapping", "lightweight research", "too much design documentation", or "get the team designing together". Also trigger when reducing design-documentation overhead, getting cross-functional teams to co-design, or running fast usability experiments. Covers hypothesis statements, MVPs for UX, and cross-functional collaboration. For Build-Measure-Learn, see lean-startup. For usability audits, see ux-heuristics.' license: MIT metadata: author: wondelai version: "1.4.0"
Lean UX Framework
A practice-driven approach to UX that replaces heavy deliverables with rapid experimentation, cross-functional collaboration, and continuous learning. Lean UX shifts the question from "What should we design?" to "What do we need to learn?"
Core Principle
Outcomes over outputs. The value of a design is measured not by the fidelity of the deliverable but by the change in user behavior it produces.
The foundation: Traditional UX waterfalls requirements into wireframes, mockups, specs, and code—losing context and hiding untested assumptions at every handoff. Lean UX compresses the distance between idea and evidence: declare assumptions, form hypotheses, run the smallest possible experiment, and let real user behavior settle the argument. Shared understanding replaces documentation; learning velocity replaces pixel perfection.
Scoring
Goal: 10/10. Score a UX process, design plan, or team workflow by the eight-row Quick Diagnostic below: award ~1.25 points per row answered "yes" (8 yeses = 10). Bands:
- 9-10 — assumptions declared, hypotheses with pre-committed success criteria, lowest-fidelity experiments, whole-team design, weekly research, outcome (not output) metrics, dual-track agile, and a recently invalidated hypothesis on the books.
- 5-6 — hypotheses exist but criteria are vague or fidelity is over-invested; design and research still partly siloed.
- <=3 — heavy deliverables, untested assumptions, output-counting, no experiment log.
Always state the current score, the diagnostic rows that failed, and the specific fix for each.
Framework
1. Declaring Assumptions
Core concept: Every design starts with assumptions. Lean UX makes them explicit so they can be prioritized and tested, rather than baked invisibly into specifications.
Why it works: Unspoken assumptions mean teams build on shaky ground and discover problems only after launch; surfacing them early focuses energy on the riskiest ones and reduces the cost of being wrong.
Key insights:
- Business assumptions define what must be true for the business (revenue model, market size, willingness to pay); user assumptions define who users are and how they behave
- Prioritize on two axes: risk (how damaging if wrong) and uncertainty (how little we know)
- Test high-risk, high-uncertainty assumptions first
- Write assumptions collaboratively as a team, not in isolation
Product applications:
| Context | Application | Example | |---------|-------------|---------| | New feature kick-off | Assumption mapping workshop | "We assume users want to share reports with teammates" | | Roadmap planning | Rank features by assumption risk | Prioritize features whose success depends on untested beliefs | | Stakeholder alignment | Expose hidden assumptions across roles | PM assumes pricing works; engineer assumes scale; designer assumes flow |
Ethical boundary: Assumptions must be honest assessments, not post-hoc justifications—if leadership has already committed to a direction, acknowledge the constraint rather than pretending it's open to falsification.
See references/hypothesis-canvas.md when running an assumption workshop or writing a hypothesis — the risk/uncertainty prioritization matrix, business-vs-user assumption split, and fillable hypothesis and sub-hypothesis templates.
2. Hypothesis Statements
Core concept: A hypothesis translates an assumption into a testable prediction, linking a proposed change to a measurable outcome for a specific user segment.
Why it works: Hypotheses force precision—instead of "make onboarding better," the team commits to a prediction that can be proven or disproven, which prevents scope creep and makes the learn step unambiguous.
Key insights:
- Standard format: "We believe [outcome] will happen if [persona] achieves [action] with [feature]"
- Every hypothesis specifies persona, action, outcome, and measurable signal
- Sub-hypotheses break a large bet into independently testable parts
- Agree on what "validated" and "invalidated" look like before running the experiment
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Feature design | Write hypothesis before wireframing | "We believe trial-to-paid conversion will rise 10% if new users complete a guided setup wizard" | | A/B tests | Formalize test rationale | "We believe click-through will rise 15% if we move the CTA above the fold" | | Sprint planning | Attach hypothesis to each story | Story: "filter by date." Hypothesis: "task completion time drops 30%" |
Ethical boundary: Never cherry-pick metrics after the fact to declare a hypothesis validated—pre-commit to success criteria.
See references/outcome-metrics.md when picking the measurable signal for a hypothesis or defining team success — outcomes-vs-outputs, leading-vs-lagging indicator pairs, UX OKRs, and the vanity metrics to avoid.
3. MVPs and Experiments
Core concept: An MVP in Lean UX is the smallest design artifact that can test a hypothesis with real users—a learning tool, not a product launch.
Why it works: A paper prototype tested with five users in a hallway can invalidate a hypothesis that would otherwise consume a full engineering sprint; matching experiment fidelity to assumption risk maximizes learning per unit of effort.
Key insights:
- Experiments range from low fidelity (paper prototypes, concierge tests) to high fidelity (coded A/B tests, Wizard of Oz)
- Choose the lowest-fidelity experiment that can answer the question
- A good experiment has a clear hypothesis, defined audience, measurable signal, and time box
- Proto-personas can stand in for full research when speed matters, but must be validated later
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Early concept validation | Paper prototype or clickable mockup | Sketch 3 concepts, test with 5 users same day | | Demand validation | Landing page smoke test | "Sign up for early access" measures real interest | | Usability validation | Clickable prototype test | Figma prototype tested with 5-8 users | | Pricing validation | Painted door test | Show pricing page, measure click-through before building billing |
Ethical boundary: Smoke tests and fake doors must not mislead users into believing a product exists—disclose test status and offer an opt-out.
See references/experiment-patterns.md when choosing or designing an experiment — the full catalog of experiment types with when/when-NOT-to-run notes, the experiment selection matrix and fidelity ladder, and a design template.
4. Collaborative Design
Core concept: Design is a team sport. Lean UX replaces the solitary designer-then-handoff model with cross-functional sessions where developers, PMs, and designers sketch solutions together.
Why it works: Developers who helped sketch the solution don't need a 40-page spec to build it—shared understanding replaces documentation, diverse perspectives generate more creative solutions, and handoff waste drops dramatically.
Key insights:
- Design Studio method: diverge (individual sketching), present, critique, converge (refined sketch), iterate
- The goal is informed commitment, not consensus: the team agrees on what to test, not what is "right"
- Cross-functional means engineers, QA, data analysts, and stakeholders sketch too
- Style guides and pattern libraries are living documents; reduce deliverables to the minimum needed for shared understanding (often a whiteboard photo)
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Sprint kick-off | Design Studio session (90 minutes) | Whole team sketches solutions to the sprint's hypothesis | | Feature exploration | Collaborative sketching workshop | 6-up sketches: each person draws 6 ideas in 5 minutes | | Remote teams | Virtual whiteboard sessions | FigJam or Miro board with timed sketch rounds |
Ethical boundary: Collaboration must not become design by committee—a designated designer synthesizes input; the team does not vote on pixels.
See references/collaborative-design.md when facilitating a Design Studio — the step-by-step workshop protocol (timings, materials, remote variants) and how to keep style guides as living documents.
5. Feedback and Research
Core concept: Continuous, lightweight research replaces big-bang usability studies—small research activities embedded in every sprint instead of quarterly reports.
Why it works: Findings only change a decision while it is still cheap to reverse, so research value decays with every sprint between learning and the decision it informs; small weekly studies keep that gap near zero, which a quarterly report never can.
Key insights:
- Research types: usability tests, customer interviews, A/B tests, analytics review, surveys, diary studies
- Five users uncover approximately 85% of usability problems (Nielsen)
- Continuous cadence: recruit weekly, test weekly, synthesize weekly
- The whole team should observe at least some sessions to build empathy
- Proto-personas are refined and eventually replaced by evidence-based personas
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Weekly usability testing | Test prototype with 3-5 users every Thursday | "Testing Thursday" ritual with rotating facilitators | | Post-launch learning | Monitor analytics + 3 follow-up interviews | Find drop-off points, interview churned users | | Persona validation | Compare proto-persona assumptions to interview data | "We assumed power users are marketers; data shows ops managers" |
Ethical boundary: Conduct research with informed consent—participants should understand how their data is used and be free to withdraw.
6. Integration with Agile
Core concept: Lean UX works inside Agile via dual-track development: discovery (learning what to build) and delivery (building it) run in parallel.
Why it works: Design work doesn't fit neatly into a delivery sprint; running discovery one sprint ahead means validated designs are ready when the delivery sprint begins, instead of design forever catching up.
Key insights:
- The discovery track (research + design) feeds the delivery track (engineering + QA), staggered one sprint ahead
- User stories gain a hypothesis and success metric alongside acceptance criteria
- "Definition of Done" for UX includes validated learning, not just shipped pixels
- Backlog items from invalidated hypotheses are removed, not deferred
Product applications:
| Context | Application | Example | |---------|-------------|---------| | Sprint planning | Include hypothesis validation in sprint goals | "Sprint goal: validate that inline editing cuts task time 20%" | | Backlog refinement | Attach experiment results to stories | Story moves to delivery only after hypothesis is validated | | **Ret
Truncated for display — read the full file on GitHub.
Related Skills
taste-skill
116.8kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Agent-Reach
112.3kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
79.8kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
Claude-Code-Game-Studios
39.7kTurn Claude Code into a full game dev studio — 49 AI agents, 72 workflow skills, and a complete coordination system mirroring real studio hierarchy.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
