Skill UI

This skill processes raw, report-only optimization observations generated by Caveman tools. It guides the user through an evidence-first, rigorous evaluation process, requiring explicit operator choice, designing a minimal candidate change, and performing a paired baseline evaluation using identical fixed inputs. It ensures that any proposed code change is validated against quality checks and established performance metrics before implementation.

A read-only skill designed for deep analysis of Itô baskets and prediction market data. It allows users to index live basket catalogs, perform deterministic gap analysis by comparing a basket against user research or a watchlist, generate source-grounded market intelligence briefs, or draft non-executable planning worksheets. This tool is strictly observational and never advises, orders, or executes trades.

A skill for analyzing and creating literature in Lao She's Beijing-flavored style, focusing on citizen life, humor and pathos, dramatic dialogue, and plain description. Ideal for writing stories, essays, or scripts with vernacular Beijing language and social observation.

A methodology for creating deep, technical AI content with a cool, anti-hype perspective. Covers research paper analysis, industry trend observation, and sharp commentary on AI events, emphasizing plain-language translations of technical reports.

A comprehensive guide and set of rules for writing effective, warm, and actionable comments across various development platforms. It helps users structure feedback for GitHub PRs, issue trackers, or Slack messages, ensuring the tone is consistently thoughtful, direct, and focused on clear next steps, rather than simply listing observations.

A set of behavioral guidelines derived from Andrej Karpathy's observations to help users writing, reviewing, or refactoring code with Large Language Models (LLMs). It guides developers to avoid overcomplication, make surgical changes, explicitly surface assumptions, and define verifiable success criteria, promoting clean, simple, and robust code development.

Techniques for extending effective context capacity via compaction, observation masking, KV-cache optimization, and partitioning to reduce tokens, cost, and latency in agent systems.

Defensive guide for reviewing authorized EDR telemetry and offline forensic evidence, classifying hook and evasion observations, and assessing detection coverage without operational bypass steps.

Locate the root cause of a confirmed failure (stack trace, failing assertion, error message, or wrong-value observation) using backward-then-forward semi-formal tracing. Provides a structured report with Primary Fault, Fault Confidence, and minimal remedy.

Retrieve NOAA NEXRAD radar and GOES satellite products from public cloud archives, select scans by observation time, and validate downloaded scientific files.

A governance skill designed to secure AI agent tool calls using advanced authorization policies. It implements Cedar policies to control which tools agents can access and under what conditions. Key features include running in 'Shadow mode' for safe observation, 'Enforce mode' for strict blocking, and generating cryptographically signed Ed25519 receipts to provide an immutable, tamper-evident audit trail for every policy decision.

Analyze weather radar and satellite observations as remote-sensing data: validate product metadata, geometry and calibration, derive storm and cloud structures, track their evolution, and quantify uncertainty for short-term nowcasting.