Skill UI

Analyzes a Power Pages site project's source code and dependencies for various security issues. It runs static analysis (opengrep) and dependency scanning (trivy) to detect vulnerable packages, hard-coded secrets, license compliance problems, and general code patterns. Use this skill to audit code safety, check dependencies, or perform a comprehensive security audit.

This skill provides an end-to-end wizard for creating a Dataverse Mobile Offline Profile, enabling Power Apps to function in an offline environment. It meticulously designs per-table row scopes, manages relationships, and defines column synchronization frequency through a multi-gate approval flow. It creates necessary Dataverse entities and configures the full app-level offline metadata, ensuring seamless offline data access for native mobile clients (iOS/Android).

This guide teaches the fundamental concepts of real-time computer graphics programming, including the vertex-fragment pipeline, coordinate space transformations, and UV mathematics. It covers implementing common visual effects (like dissolve, rim light, and scrolling textures) using portable GLSL and HLSL equivalents, providing a strong foundation for game development shaders.

This internal deterministic quality gate is used before meta-short-drama final publication. It rigorously verifies the media content's integrity by checking real-provider image/video receipts, confirming paid submission dispositions, and utilizing ffprobe to validate decodability and expected final durations. The tool is designed to sanitize all sensitive data, ensuring only bounded, auditable information is passed to the next stage.

A content creation framework inspired by Chinese AI media creator Kazik (数字生命卡兹克). Covers data-driven storytelling, emotional pacing, self-deprecating humor, AI product reviews, and scalable workflows for producing impactful AI tech articles, videos, and event coverage.

skill-doctor grades the performance and quality of an AI agent setup using real, local conversation history (e.g., Claude Code/Codex sessions). It scores skills against defined rubrics for efficiency and code quality, drafts evidence-backed suggestions for improvement, and generates a comprehensive, shareable local report. All processing is strictly local and private.

This tool guides the analysis of work sessions by detecting user feedback (corrections, successes, edge cases). It evaluates these signals against quality criteria (Complete, Precise, Atomic, Stable) to propose structured improvements, update skill documentation (SKILL.md), and maintain a changelog.

This skill automates the final deployment phase of an agent system. It converts a validated build sheet into four ordered JSON payloads (environment, agent, session, kickoff) and generates a resumable `launch.sh` script. Crucially, this process enforces API-key safety by requiring the user to run the script in their shell, pulling the Anthropic API key from the environment variable, thus ensuring the key never appears in logs or files. It provides a structured, multi-step launch sequence.

A disciplined methodology for fixing bugs and small behavioral changes with minimal impact. It involves tracing symptoms to the narrowest responsible mechanism, implementing changes only at that layer, and adding highly focused regression proofs to ensure surrounding functionality remains intact. Ideal for critical systems requiring maximum stability.

This advanced skill stress-tests legal submissions, motions, or arguments by generating an adversarial critique that survives multiple layers of challenge. It goes beyond superficial attacks, ensuring every point is load-bearing and defensible against a single-sentence reply. Ideal for red-teaming, stress-testing, or refining arguments to withstand rigorous scrutiny.

This skill guides the extraction of structured data from multiple documents (PDF, DOCX) into a comprehensive, cited Excel matrix. It is ideal for legal, contractual, or research review processes where users need to compare specific clauses, provisions, or pieces of information across numerous source files, ensuring every extracted data point includes a citation (page number and context quote) for verifiable accuracy.

Distill video taste interviews into validated style packs, apply measured cadence to local footage, and export editable EDL/FCPXML timelines with provenance audits. All operations are local, deterministic, and offline.