Open-source agent skills for founders building products that actually sell.
These are Agent Skills — structured instruction files that give AI agents (Claude, Cursor, Copilot, and others) deep, opinionated expertise in specific domains. This collection covers the decisions that most often determine whether a software product succeeds or dies — from idea validation to code quality to conversion.
| Skill | What it does |
|---|---|
| pmf-advisor | Validates ideas, prioritizes roadmaps, and assesses PMF using RICE, JTBD, Value Proposition Canvas, and the Sean Ellis Test |
| icp-discovery | Destroys vague "anyone who..." ICPs through 6 stages of iterative questioning — behavioral, situational, economic, and channel profiling |
| pricing-strategy | Finds the right value metric, pricing model, and price point using Van Westendorp, WTP interviews, and tier structuring frameworks |
| qa-advisor | Audits codebases across test quality, maintainability, security (OWASP Top 10), reliability, and DORA metrics — produces a structured QA Audit Report |
| website-icp-fit | Evaluates whether a website's copy, messaging hierarchy, and funnel structure are correctly aligned to convert the defined ICP |
| high-ticket-conversion-architect | Designs and audits landing pages for high-ticket services ($10k+) using Intent-Friction Ratio, Constructive Friction, Shadow Proof, and the 7 Messaging Blocks |
Built by Wavect GmbH — "More than Engineers. Build to Sell."
# Install a single skill for the current project
cp -r pmf-advisor .claude/skills/
# Or install all six
for skill in pmf-advisor icp-discovery pricing-strategy qa-advisor website-icp-fit high-ticket-conversion-architect; do
mkdir -p .claude/skills/$skill
curl -sL https://raw.githubusercontent.com/wavect/ai-skills/main/$skill/SKILL.md \
-o .claude/skills/$skill/SKILL.md
doneThen in Claude Code:
/skills pmf-advisor
Copy the SKILL.md file from any skill directory into your project's
.claude/skills/ folder (Claude Code) or wherever your agent loads skill
context from. The SKILL.md format is tool-agnostic.
Point your agent at the raw GitHub URL:
https://raw.githubusercontent.com/wavect/ai-skills/main/pmf-advisor/SKILL.md
https://raw.githubusercontent.com/wavect/ai-skills/main/icp-discovery/SKILL.md
https://raw.githubusercontent.com/wavect/ai-skills/main/pricing-strategy/SKILL.md
https://raw.githubusercontent.com/wavect/ai-skills/main/qa-advisor/SKILL.md
https://raw.githubusercontent.com/wavect/ai-skills/main/website-icp-fit/SKILL.md
https://raw.githubusercontent.com/wavect/ai-skills/main/high-ticket-conversion-architect/SKILL.md
"Most software projects don't lack engineers, they lack product experience."
Activates when you need to validate an idea, prioritize a backlog, or figure out why traction isn't converting into retention.
Frameworks included:
- RICE Scoring — prioritize roadmap items by evidence, not gut feeling
- Jobs-to-Be-Done — surface the emotional job behind the functional request
- Value Proposition Canvas — map pains/gains to your actual offering
- Sean Ellis Test — the 40% threshold that separates PMF from delusion
Output: PMF Risk Assessment → Recommended Framework → Diagnostic Questions → Next Step
"You cannot market to everyone. Everyone is no one."
The most misused term in early-stage startups is "ICP." This skill will not let you get away with "SMBs," "SaaS companies," or "people who need X." It works through six stages until you can picture one specific human at their desk with a specific problem.
Frameworks included:
- YC / Michael Seibel — "Hair on Fire" customer identification
- Steve Blank — behavior over demographics (work backwards from who paid)
- Rob Fitzpatrick — Mom Test applied to ICP qualification
- Peter Thiel — dominate the smallest market first, then expand
- April Dunford — position for the customer where you are the obvious choice
Output: ICP Card — role, triggering situation, burning problem, behavior today, economic profile, top 3 channels, riskiest assumption, beachhead size estimate
"Your price is a signal. Undercharging says you don't believe in your own value."
Most founders underprice by 2–5x. This skill starts by identifying the right value metric (the unit pricing should scale on), then picks the right model, then finds the right number — in that order. It never starts with the number.
Frameworks included:
- Patrick Campbell — value metric identification (the most leveraged pricing decision)
- Van Westendorp Price Sensitivity Meter — 4-question method to find acceptable price range
- Willingness-to-Pay interviews — how to ask without anchoring the answer
- Simon-Kucher pricing waterfall — discount and packaging discipline
- Tier structure principles — why 3 tiers, what goes where, upgrade trigger design
Output: Pricing Strategy Card — value metric, recommended model, price range, tier structure, upgrade trigger feature, biggest pricing risk, next experiment
"Coverage is vanity. Meaningful tests are sanity."
Most CI pipelines have a green badge and a codebase that will still fail in production. This skill audits five dimensions: test quality, maintainability, security, reliability/scalability, and delivery health — and it does not soften findings.
Frameworks included:
- Martin Fowler's test double taxonomy — dummy, stub, spy, mock, fake (and when each is correct)
- London vs. Chicago school TDD — detecting accidental mixing and its consequences
- OWASP Top 10 — IDOR, JWT
alg:none, SSRF, mass assignment, timing attacks, and more - Property-based testing — invariants, round-trip, and oracle properties via fast-check / Hypothesis
- Contract testing (Pact) — preventing silent API breakage in microservices
- DORA metrics — deployment frequency, lead time, MTTR, and change failure rate as quality signals
- Hexagonal architecture — detecting infrastructure entanglement that makes testing impossible
- Load testing methodology — ramp, spike, soak, stress, breakpoint, and capacity patterns
Output: QA Audit Report → Critical / High / Medium findings (file:line, specific remediation) → Test Quality Scorecard → Security Posture → DORA Assessment → Architectural Testability → Immediate Action → Three-Month Roadmap
"Your homepage is not for everyone. If it tries to speak to everyone, it converts no one."
Traffic without conversions is a messaging problem, not a product problem. This skill evaluates a website — by URL or source code — against a defined Ideal Customer Profile. It refuses to produce recommendations without a specified ICP, because a homepage audit without an ICP is interior design critique.
Frameworks included:
- Eugene Schwartz's 5 Levels of Awareness — diagnosing the mismatch between traffic source and page message
- StoryBrand SB7 — detecting when the company is positioned as the hero instead of the guide
- April Dunford's Positioning — extracting (or exposing the absence of) a clear positioning statement
- Cialdini's 7 Influence Principles — auditing social proof ICP match, authority signals, reciprocity offers
- Fogg Behavior Model — identifying Ability failures (form length, hidden pricing, unclear CTA) over Motivation failures
- The 4 U headline formula — Useful, Urgent, Unique, Ultra-specific applied to every H1
- Funnel architecture — the correct scrolling sequence and what most homepages get wrong
Output: ICP-Website Fit Audit Report → 5-Second Test → Critical / High / Medium findings → Messaging Scorecard → Psychological Triggers Audit → Funnel Structure → Trust Architecture → Pricing Page → Anti-Patterns → The One Change That Would Move Conversion Most → 90-Day Roadmap
"Volume-based optimization is for $99 products. At $10k+, you are not optimizing clicks — you are bridging a trust gap."
For premium B2B and luxury service offers above $10,000, conventional landing-page best practices invert. Lower friction reduces conversion. Generic social proof reduces conversion. "Book a Call" reduces conversion. This skill designs and audits high-ticket pages on the principles of Relationship Selling, Risk Reversal, and Selective Enrollment — engineering the page so the right buyer self-qualifies and the wrong buyer self-disqualifies.
Frameworks included:
- Intent-Friction Ratio — when to add Constructive Friction and when to remove Maladaptive Friction
- The 7 Messaging Blocks — Headline, Context, Mechanism, Proof, Selective Enrollment, CTA, Ethics
- Proof of Work + Shadow Proof — uncomfortably specific operational detail that fraudsters cannot fabricate
- Selective Enrollment / Disqualification — repelling wrong-fit buyers as a conversion lever
- The ROI Bridge, Decision Frameworks, Response Velocity Signals
- Confidently Boring aesthetic, Von Restorff Effect, Decoy Effect, Ben Franklin Effect, Loss Aversion framing
- Traffic-temperature calibration (Cold / Warm / Hot) to dictate persuasion length
Output: High-Ticket Landing Page Architecture → Intent-Friction Assessment → 7 Messaging Blocks (with copy and proof) → Specialized Modules → Psychological Heuristics → Aesthetic Direction → The One Change That Would Convert Most → 90-Day Deployment Roadmap
This repository publishes a machine-readable skill index at:
https://raw.githubusercontent.com/wavect/ai-skills/main/index.json
The index follows the Agent Skills Discovery RFC v0.2.0 and includes SHA-256 digests for each skill file so agents can verify integrity.
For sites that use the .well-known discovery pattern (e.g. isitagentready.com),
Wavect's skills are also published at:
https://wavect.io/.well-known/agent-skills/index.json
Wavect GmbH is an Austrian software product studio that helps startups and corporates hit Product-Market Fit. We combine technical execution with product strategy — the two things most agencies pick one of.
Services:
- Fractional Co-Founders — strategic + technical co-founder, 0% equity, from €400/week
- Software Development — full-stack from MVP to scale (Web, Mobile, Web3, AI)
- Software Quality Assurance — manual + automated testing, from €750/week
These skills are our methodology made public. We use the same frameworks in client engagements. If you want a sparring partner who has done this with real products, book a free call.
See CONTRIBUTING.md. In short: corrections, sharper
questions, and new product/go-to-market skills in the same spirit are welcome.
When editing a SKILL.md, regenerate the digest and update index.json:
shasum -a 256 skill-name/SKILL.md
# then update the digest field in index.jsonMIT — see LICENSE. Use, fork, and adapt freely. Attribution appreciated but not required.