How to use
- Copy the installation command and review the source package before installing it in your own agent environment.
- Check the source document for required tools, credentials and supported hosts.
- Describe your input and desired deliverable; review the output against the source material.
npx skills add https://github.com/calionauta/stelow --skill stelow-product-discoveryBefore you start
- Installation does not include a model subscription, API allowance or permission to access your files.
- Repository stars and installation counts describe activity, not an independent quality or security test.
Rendered from the source repository. Headings, examples, code, tables, links, and referenced images are preserved.
Product with Short Learning Cycles
This guide is a method to replace speculation with evidence, step by step. Core principle: experiment before building. Reduce uncertainty with small, fast, and cheap experiments.
"Life is too short to build something that nobody wants." — Ash Maurya
Method Structure
The process has 8 stages (not necessarily linear):
- Find and understand the audience
- Define the market
- Define and prioritize solutions
- Develop and evaluate the offer
- Assess commitment
- Discover the process
- Evolve to a viable business
- The pulse of the system: signals in tension
→ Read reference files as needed:
references/01-experimentation-principles.md— criteria for reliable data + 6 principlesreferences/02-stages-1-to-3.md— audience, market, solutionsreferences/03-stages-4-to-6.md— offer, commitment, manual processreferences/04-stage-7.md— viable business, product-market fit, channelsreferences/05-stage-8-signals.md— pulse of the system, signals in tensionreferences/06-appendix-business-models.md— business models, open source, pricingreferences/07-appendix-strategies.md— marketplace, solutions, trust, bonuses, promotions, ads, lifecycle
When to Read Each Reference
| User asks about... | Read... |
|---|---|
| How to validate an idea / experiment | 01, 02 |
| Target audience, market, jobs to be done | 02 |
| Solution, product, features | 02, 03 |
| Offer, advertisement, early adopters | 03 |
| Pre-sale, commitment, waitlist | 03 |
| MVP, manual process, concierge, wizard of oz | 03 |
| Product-market fit, channels, viable business | 04 |
| Metrics, retention, growth | 04, 05 |
| Pricing, pricing models | 06 |
| Business models, revenue, costs | 06 |
| Open source as strategy | 06 |
| Marketplace, supply and demand | 07 |
| Trust, guarantees, social proof | 07 |
| Bonuses, launch promotions | 07 |
| Online ads, stages of awareness | 07 |
| Product lifecycle, innovation | 07 |
Interaction Tool Guidelines
IMPORTANT: When the user needs to choose between predefined options, ALWAYS use the question tool (if available) with enumerated format:
- Options with short
labelanddescription - Examples: topic selection, next steps, etc.
When question tool is not available, use enumerated text in chat (A/B/C/D or 1/2/3).
Note for this skill: This is primarily a reference/exploratory skill. If the user wants to explore a specific concept, you may ask clarifying questions to understand context before recommending which references to read or which stage to focus on.
Rules for Using This Skill
- Never omit details — the e-book is already concise; every detail matters.
- Use concrete examples from the method when explaining concepts.
- Reference specific tools and tactics mentioned in the method.
- Preserve the order and logic of stages when guiding the user.
- Combine stages when context demands (e.g., stages 4 and 5 often go together).
Attribution
This skill is based on the e-book "Guia para criar produtos com aprendizado em ciclos curtos" (Guide to Creating Products with Short Learning Cycles) by calionauta.
Author: calionauta Substack: Practical Guide to Validate Ideas GitHub: @calionauta
This skill was translated and adapted by permission and attribution to the original author.
Original Repository: github.com/calionauta/my-opencode-config
Extended Deep-Dives (Optional)
For specialized analysis, these standalone skills are vendored in this repo under skills/ (historically published at github.com/calionauta/agent-sync-public):
| Skill | Description | Lines |
|---|---|---|
stelow-product-health | Product health monitoring via signals in tension | ~100 |
stelow-product-pricing | Pricing strategy: Exchange Base, Consumption Control, Interest Alignment, Trust Pact | ~100 |
stelow-product-open-source | OSS as business strategy: Proprietary Problem, Org Structures, Competition Protection | ~60 |
stelow-product-trust-building | 10 pillars of trust, guarantees, perception, commitment | ~60 |
stelow-product-promotions | MAGIC framework, 4 launch strategies, bonuses | ~60 |
stelow-product-business-models | Cost reduction & revenue generation models | ~80 |
stelow-product-ads | 5 awareness stages, ad categories | ~50 |
stelow-product-paywall | Paywall & onboarding monetization funnel (paywall-first build order, 3 funnel benchmarks, trial policy, web2app) | ~310 |
Install: already vendored in this repo — no separate install needed. Note: These are optional enhancements. The core method works without them — the references/ contain all the necessary content.
Examples
Example: validate a meal-planning app idea in two weeks
Input: "I have an idea for an app. What do I do first?"
Steps:
- Frame the riskiest assumption (will busy parents pay for planning?).
- Pick the shortest-cycle experiment (concierge week, not a build).
- Define abort criteria before running.
Output: Experiment plan with success/abort thresholds and next step either way.
Edge Cases
User wants to build immediately
- Slow down explicitly: name the unvalidated assumption building would bet on.
No access to users
- Proxy with adjacent audiences; mark evidence weaker.
Experiment succeeds ambiguously
- Treat as not-validated; tighten the threshold and re-run once.
Entry (mode detection)
When this skill loads, check for the stelow workflow marker:
if [ -n "$STELOW_WORKFLOW" ] && [ -n "$STELOW_STATE" ]; then
echo "stelow: workflow mode (state=$STELOW_STATE)"
else
echo "stelow: standalone mode (no STELOW_WORKFLOW marker)"
fiIn standalone mode (no marker), run the existing skill body unchanged. In workflow mode, skip to ### Workflow slice and emit a complete ## Hand-off (workflow mode) block at the end. See references/host-levers.md for the full marker protocol (SCOPE-9).
Hand-off (workflow mode)
stage : context
description : Strategic context. Market analysis, JTBD, domain detection. Gated by `context:5` (appetite/review mode): Product Spec Ga
status : <done|partial|blocked>
artifacts : <paths created or modified>
next-candidate : shape
gate : none
rework-on : setupWorkflow mode: emit the above Hand-off block verbatim, then stop. The router skill consumes the next-candidate field and calls scripts/stelow advance <next-candidate> to move state forward.
Workflow slice
Workflow mode for the context stage. Standalone behavior lives in the rest of this file (unchanged). Summary:
Strategic context. Market analysis, JTBD, domain detection. Gated by context:5 (appetite/review mode): Product Spec Gate+Auto skips; others use reduced ask. SPrimary actions (per stages.yaml): read, write. Run only the actions that produce the artifacts promised in ## Hand-off; skip anything that does not advance the workflow.



