Skip to content

Instantly share code, notes, and snippets.

View w1ne's full-sized avatar
πŸš€

Andrii Shylenko w1ne

πŸš€
View GitHub Profile
@w1ne
w1ne / SKILL.md
Created April 1, 2026 21:00
Risk Storm β€” 5 risk lenses + pre-mortem synthesis for product decisions
name risk-storm
description 5 risk lenses attack your product, then pre-mortem synthesis reveals how it dies. Use when: risk storm, what could go wrong, pre-mortem, is this risky, what am I missing, stress test this idea.
argument-hint [your product, feature, or launch plan]

Risk Storm β€” Product Risk Identification

Combines risk storming, pre-mortem analysis (Gary Klein), and FMEA scoring into one skill. 5 risk lenses attack your product from different angles, then a pre-mortem synthesis reveals the most likely causes of failure.

@w1ne
w1ne / SKILL.md
Created March 31, 2026 11:51
Council Skill β€” Karpathy-style multi-perspective decision advisor for Claude Code
name council
description Run a multi-perspective advisory council on any decision. Spawns 5 advisors with distinct thinking styles (Contrarian, First Principles, Expansionist, Outsider, Executor), runs anonymous peer review, and produces a chairman synthesis with a clear recommendation. Based on Karpathy's LLM Council method. Triggers on "council this", "council", "I need multiple perspectives", "should I", "help me decide", "I keep going back and forth", or any high-stakes decision where the user wants structured disagreement rather than a single agreeable answer.

Council β€” Multi-Perspective Decision Advisor

Based on Andrej Karpathy's LLM Council method. Instead of polling multiple models, this skill forces 5 distinct thinking styles onto the same question using parallel subagents, then runs anonymous peer review to catch what no individual advisor sees.

Use when: The cost of being wrong is high and you keep going back and forth. Skip this for simple questions β€” it's designed for real deci

@w1ne
w1ne / capabilities.txt
Created March 9, 2025 23:47 — forked from jlia0/agent loop
Manus tools and prompts
# Manus AI Assistant Capabilities
## Overview
I am an AI assistant designed to help users with a wide range of tasks using various tools and capabilities. This document provides a more detailed overview of what I can do while respecting proprietary information boundaries.
## General Capabilities
### Information Processing
- Answering questions on diverse topics using available information
- Conducting research through web searches and data analysis
### Keybase proof
I hereby claim:
* I am w1ne on github.
* I am shylenko (https://keybase.io/shylenko) on keybase.
* I have a public key ASCaPJK9VZLGldrXejv_V5Ee0jbpIrPfjo9-o_WETKF8Xgo
To claim this, I am signing this object: