From one interview guide to a project you can resume
Wondelai Skills
65 Methods and Project Workflows
Wondel.ai packages product, design, marketing, management, and engineering methods into 51 skills and 14 guided journeys. Choose a concrete problem before committing to a longer workflow.
Hosts listed by the author
- Claude Code
- Codex
- Cursor
- OpenClaw
- Hermes
- GitHub stars
- 2,536
- Version
- 1.8.0
- Checked
Use one method to finish something you can check
Start with an interview guide or homepage message: the scope is small and missing inputs become visible. Choose create-business or improve-website when you need decisions across methods and stages, and keep the project documents.
Pick a task and get started
Open Claude Code in the target project, select a task, and enter its installation commands one at a time. Confirm the required skills are loaded, prepare real inputs, and send the prompt. Different tasks require different collections.
Turn a customer hypothesis and draft questions into a conversation about recent events. Start here when you need evidence that the problem exists.
Install the matching collections in Claude Code
Enter these commands one at a time in a Claude Code session, not a shell. Skip the first line if the marketplace is already added. Commands install the selected task’s collections at the version available then. Journey commands cover the stages shown here; check the required skills again before later stages.
Read the author’s setup guide/plugin marketplace add wondelai/skills /plugin install product-strategy@wondelai-skillsFirst path: prepare a customer interview guide
Select the first task. In your project, create brief.md with the audience, problem, decision, and known facts, plus questions-draft.md with your current questions or a note that none exist. In the Claude Code session, add the marketplace and install product-strategy one command at a time; skip the first line if the marketplace is already added. Ask the assistant to confirm it can read mom-test and its references, then send the task prompt. Expect docs/interview-guide.md and a note template. Check that questions ask about concrete events without suggesting a product answer, then conduct the real interviews.
Read the author’s Claude Code installation guideGive your AI this first task
These prompts are editorial examples based on the author’s workflows. Filenames are replaceable input and output conventions. Apart from the journeys’ prescribed PLAN and shared artifacts, they are not files supplied by installation.
Use mom-test from the product-strategy collection. Read brief.md and questions-draft.md to prepare interviews about how freelancers handle late client payments. The brief should name the customer group, decision to make, and known facts; clarify missing essentials. Produce 10–12 main questions with follow-ups about a recent incident, existing alternatives, and actual time or money spent. Remove pitches, implied compliments, and hypothetical adoption questions. Save docs/interview-guide.md and docs/interview-notes-template.md. Separate quotes, facts, interpretations, unknowns, proposed commitments, and fulfilled commitments. Review the questions against the skill. No interviews have happened yet: do not report observed talk ratios, customer commitments, or validated demand. Finish with the interviews I need to conduct and the evidence to bring back.
Check what you receive
- An interview guide and note template that follow concrete events and real alternatives.
- Check each question for leading language. Analyze facts, contradictions, and commitments after actual interviews; a prepared guide is not validation.
More about this skill
Wondelai Skills is the method library in wondelai/skills. The reviewed revision contains 65 distinct skill directories: 51 frameworks and 14 orchestration workflows. It gives an open-ended request a sequence of questions, diagnostic checks, and a deliverable structure for product, marketing, and engineering work.
The source methods have different authors. mom-test adapts Rob Fitzpatrick’s The Mom Test; storybrand-messaging uses Donald Miller’s StoryBrand. technical-documentation independently adapts Google’s documentation guidance and adds Keep a Changelog. Wondel.ai packages the skills and authors the journeys; individual entries credit the original methods so you can continue with the books or guidance.
A single skill addresses one task: turn leading questions into behavioral interviews, clarify a homepage, or audit a README. A journey connects several skills. create-business has ten stages, from the customer’s job through interviews, prototypes, product definition, strategy, positioning, offers, and pricing. improve-website has eight, moving from conversion evidence into usability, visual design, messaging, and performance.
Journeys resume through project files. Each keeps its own PLAN document while customer, positioning, metrics, and other artifacts share agreed locations under docs/. A later session reads the existing material before choosing a stage. Decisions and next actions are recorded at stage boundaries; awaiting-evidence preserves a pause for external inputs. Keeping these files gives the next session the decisions already made.
Claude Code offers ten installable collections; other hosts have separate Skills and Agent Plugins paths. The five tasks below provide concrete starting points and matching collection commands. Installing metaskills supplies the orchestration: the current stage’s methods must be available or you must explicitly choose the workflow’s brief treatment, and the journey still needs your evidence and decisions.
Sources and verification
Sources checked
Based on full reads of the pinned README, installation manifests, seven SKILL files, and relevant references. Directory structure, task dependencies, and sources were checked statically; no third-party installation, model task, or website experiment was run.