Skill UI

This skill provides a CLI-based workflow for Android APK reverse engineering. It covers APK unpacking, Java decompilation, smali modification, repackaging, Frida dynamic hooking, and optional native .so analysis. It leverages local tools such as jadx, apktool, frida, adb, ida-reverse, and radare2. Useful for analyzing login, signature, security, root detection, or certificate verification logic in Android apps, always in an authorized context.

Learn to use SuperDoc CLI v1 to edit, query, and transform Word (.docx) documents. Covers stateful/stateless workflows, text replacement, formatting, comments, track changes, and history operations with JSON-based target addressing.

This workflow implements a bounded, iterative refinement loop for AI agents. It utilizes a required markdown rubric to grade the agent’s output in an isolated context. Based on the grader's verdict, it recommends the next move (sharpen, re-run, or promote). Once the agent passes the rubric, held-back evaluation cases are run in parallel to ensure generalization before final deployment.

This skill helps organizations determine mandatory reporting duties, receiving authorities, and precise deadlines following a security incident, data breach, or vulnerability exploitation. It synthesizes reporting requirements from major EU regimes, including GDPR, NIS2, DORA, and the Cyber Resilience Act, providing a cited map of obligations based on the entity's role and location. It determines *what* must be reported, citing official legal texts.

This sophisticated skill assists in-house legal teams with comprehensive legal risk assessment and classification. It uses a structured framework based on severity and likelihood, but critically, it enforces a rigorous three-step research methodology: adversarial searching for contradicting case law, systematic doctrinal web searches for context, and a temporal confidence check to ensure legal positions are current. It ensures that risk assessments are grounded in the most reliable and updated legal landscape.

Researches uncertain questions through user-approved web search or ChatGPT consultation, then presents sourced options and waits for explicit approval before writing code. Redacts sensitive data and never shares context, files, or browser state without consent.

A Python-based REST API skill for Azure Speech to Text, enabling simple speech recognition of short audio files (up to 60 seconds) without using the Speech SDK. It covers setup, authentication, request formats, and chunked transfer for efficient transcription.

Guide for splitting extracted text into chunks for LLM context windows or RAG ingestion. Covers chunk size, overlap, markdown/yaml/semantic chunkers, tokenizer-based sizing, and the standalone chunk command.

Build AI agents that mimic human computer interaction, including viewing screens, moving cursors, clicking buttons, and typing text. This skill covers frameworks like Anthropic's Computer Use and OpenAI's Operator, focusing on vision-based control, sandboxing, and security. Ideal for desktop automation, GUI automation, and browser agents.

This skill covers strategies for compressing conversation history in AI agent sessions to manage context window limits. It focuses on optimizing tokens-per-task rather than just tokens-per-request. Techniques include anchored iterative summarization, opaque compression, and regenerative summaries. It also addresses artifact trail integrity and evaluation methods like probe-based testing to ensure critical information like file paths and decisions are preserved during compression.

This skill covers persistent memory systems for Large Language Model (LLM) conversations. It includes implementing short-term, long-term, and entity-based memory to enhance context retention across sessions. Features include memory consolidation, retrieval strategies, and integrating memory into prompts. Suitable for building conversational AI applications using tools like Mem0, LangChain, or Redis.

This guide provides a comprehensive reference on Agent Skills, a standardized and powerful method for providing structured context, workflows, and tools to Large Language Models (LLMs). It details the anatomy of the SKILL.md file, the three-stage progressive disclosure loading process, and best practices for creating both capability (tool-based) and process (workflow-based) primitives. Use this resource when designing, reviewing, or debugging agent-driven applications to ensure reliable and context-efficient performance.