Systematic techniques for structuring LLM instructions to produce accurate, consistent, and high-quality outputs
Skills in this repository
lev-os/agents - Page 70
SkillsMP has collected 2,887 skills from lev-os/agents. Open a skill to review its source and details.
lev-os/agentsShowing 40 of 2,887 collected skills.
Test software properties with automatically generated inputs instead of hand-written examples, uncovering edge cases humans miss
Understand how people evaluate risk irrationally through loss aversion and reference points - predict decisions by recognizing pain of loss outweighs equivalent gain
Protocol network effects emerge when a communication or computational standard is adopted, becoming self-reinforcing as compatible products flood the market
The state of natural simplicity and wholeness before being shaped by external conditioning, maximizing power through preserving natural potential
Clarify roles when cross-functional projects cause "who's doing this?" confusion
Kim Scott's feedback framework combining caring personally and challenging directly to give kind, clear, and effective feedback
Translate ambitious product visions into strategy when suffering from feature bloat or metric obsession
Retrieval-Augmented Generation architectures that enhance LLMs with real-time external knowledge to reduce hallucinations
Clarify decision accountability by assigning five distinct roles (Recommend, Agree, Input, Decide, Perform) when navigating complex multi-stakeholder decisions
Recognize competitive environments where you must continuously evolve just to maintain position - standing still means falling behind
Build backup systems and duplicate critical components so single failures don't cascade into total system collapse
Systematically improve code structure without changing behavior using Martin Fowler's catalog of behavior-preserving transformations
Practical design tactics for developers who want to create professional UIs without formal design training
Systematic framework for designing customer referral programs with double-sided incentives, optimal timing, friction reduction, and AI-powered optimization
Comprehensive growth methodology covering loops vs funnels, activation habits, retention analysis, and multi-channel growth strategies
Extreme outcomes tend to be followed by more average ones due to statistical variation
Recognize that measurements and observations depend on frame of reference—absolute values are less meaningful than relationships and context
The First Law of Cybernetics stating that a control system must have at least as much variety (response options) as the system it aims to regulate, otherwise the environment will dominate and destabilize the system
Amplification that occurs when external input aligns with a system's natural frequency, where small well-timed efforts produce disproportionately large effects
Systematically measure and improve how frequently users return through cohort analysis and behavioral metrics to optimize for product-market fit
When your sprint/project ends - structured reflection framework that turns team experience into actionable improvements
Scoring framework using Reach, Impact, Confidence, and Effort to objectively prioritize product features and projects
Visual probability-impact assessment tool that categorizes risks on a grid by likelihood and severity to prioritize mitigation efforts
Calibrate evidence requirements proportional to claim implausibility when evaluating extraordinary assertions
Identify hub vulnerabilities when analyzing infrastructure resilience or planning targeted interventions
Predict resource needs and innovation dynamics when evaluating whether to scale up or stay small
Leverage limited resources and exclusivity to drive value, prioritization, and strategic advantage
Creating scenarios like "cybersecurity incident" or "economic downturn"—too vague to drive specific responses.
You face a continuum from acceptable to unacceptable:
German military concept meaning "point of main effort"—the decisive focal point where maximum resources, attention, and initiative must be concentrated to achieve breakthrough results.
Force proportional analysis when emotional responses fail to scale with problem magnitude
Think beyond immediate consequences to understand cascading effects and long-term impacts. Use when making high-stakes decisions, evaluating long-term strategies, or when everyone agrees on the obvious solution. Apply systematic "and then what?" questioning…
Systematic testing of how changes in input variables (assumptions) affect output results, revealing which factors drive outcomes and where model risk is concentrated.
If someone asks what's on the list immediately, you grab what's on the desk (last items) plus retrieve what you filed (first items). Middle items were neither filed nor still on the desk.
Navigate strategic uncertainty through multiple plausible future scenarios rather than single-point forecasting. Developed at Royal Dutch Shell, this framework replaces single-point predictions with 2-4 rich narratives representing genuinely different futures…
Royal Dutch Shell's pioneering scenario planning method using critical uncertainties to develop multiple plausible futures for strategic preparedness
Parallel thinking method using six colored hats to systematically explore problems from six distinct perspectives in sequence
Apply Roman Krznaric's cognitive toolkit for becoming a good ancestor through deep-time humility, legacy mindset, intergenerational justice, cathedral thinking, holistic forecasting, and transcendent goals
Add strategic bridges when teams are siloed but need rapid information diffusion