Google ADK (Agent Development Kit) architecture and core concepts. Covers agent types (LLM, Sequential, Parallel, Loop, Multi-agent), tool ecosystem, execution flows, deployment options, and observability. Model-agnostic framework for production AI agents. Use when understanding ADK architecture, selecting agent patterns, designing agent systems, or deploying to Vertex AI.
Handle comprehensive project language and user setup workflows including language selection, agent prompt configuration, user profiles, team settings, and domain selection
Control Reachy Mini robots by Pollen Robotics. Use when building apps, controlling robot movement (head, antennas, body), accessing camera/audio, or integrating with LLMs. Triggers on requests involving Reachy Mini SDK, robot control, head pose matrices, or ReachyMiniApp development.
Watch educational demo like a learner (BLIND evaluation). No test cases. No benchmark. No rubric. Honest assessment of: impression, what works, what doesn't, learner impact, recommendation. Output: agent_X_free_eval.json
Helps you create and refine new Gravito Skills. Trigger this when asked to add a new skill to the ecosystem.
Adaptive communication patterns and role selection based on user expertise level and request type (Consolidated from moai-alfred-persona-roles)
Download an image for a vocabulary card using Unsplash. Use when the user wants to add an image to a card, get an image for a word, or says "add image to [word]" or "generate image for card [id]".
Fast LLM inference with Groq API - chat, vision, audio STT/TTS, tool use. Use when: groq, fast inference, low latency, whisper, PlayAI TTS, Llama, vision API, tool calling, voice agents, real-time AI.
Real-time sentiment analysis on Twitter/X using Grok. Use when analyzing social sentiment, tracking market mood, or measuring public opinion on topics.
Optimize prompts via process goals (controllable behavioral instructions) rather than outcome goals (sparse end-result demands). Grounded in sports psychology meta-analysis showing process goals (d=1.36) vastly outperform outcome goals (d=0.09). Use when designing prompts, optimizing LLM steering, implementing CoT/decomposition patterns, or building automatic prompt optimization pipelines. Instantiates surrogate loss paradigm for discrete prompt space.
Main orchestration loop for autonomous coding. Use when running autonomous sessions, orchestrating feature completion, managing continuous loops, or coordinating agent lifecycle.
Retrieve project-specific insights from file-based memory. Use when working on features, encountering domain-specific questions, or user says "--project-recall" or "--recall" (you decide which scope, may use both). Skip for routine tasks or universal pattern questions (use coder-memory-recall). MUST be invoked using Task tool to avoid polluting main context.
自动从网络收集 Claude Code 的最新使用技巧、功能更新和最佳实践。当用户需要了解 Claude Code 的新功能、使用技巧、高级用法或 Claude Agent SDK 开发时使用此skill。支持本地缓存，避免重复搜索。
Use when user provides a simple, vague prompt and asks to expand/improve it, or when preparing prompts for another session. Triggers on "expand this prompt", "make this more specific", "improve this prompt", or similar requests.
Maintains project context and progress tracking across Claude sessions. Use at session start to load context, on session end to save progress. Triggers: "load project context", "save context", "end session", "what was I working on", "switch to [project]", "done for today". Works in both Claude Code and Claude Desktop.
Comprehensive security risk analysis for Claude skills. Use when asked to analyze security risks, review security stance, audit skills for vulnerabilities, check security before deployment, or evaluate safety of skill files. Triggers include "analyze security," "security risks," "security audit," "security review," "is this skill safe," or "check for vulnerabilities."
Copilot CLIに処理を委譲するスキル。ユーザーが「Copilotに聞いて」「Copilotで確認して」「Copilotに任せて」などと明示的に依頼したときに使用する。GitHub Copilot CLIの機能を活用して、コマンドの提案やGit操作の説明などを取得する。
AI governance and responsible AI planning including EU AI Act classification, NIST AI RMF, and AI ethics frameworks
Expert Python developer specializing in Python 3.11+ features, type annotations, and async programming patterns. This agent excels at building high-performance applications with FastAPI, leveraging modern Python syntax, and implementing comprehensive type safety across complex systems.
Implement vector search for knowledge retrieval. Use when adding RAG, semantic search, knowledge base features, or context building for the agent.
Write effective prompts for Task tool sub-agents, slash commands, and system prompts. Covers Claude 4.x prompting patterns, context engineering, output format specification, and parallel delegation. Use when spawning sub-agents, creating slash commands, or writing system prompts.
Create and manage Decision (DEC-NNN) and Assumption (ASM-NNN) records. Use when: documenting non-obvious choices, tracking risky assumptions, explaining rationale. Triggers: "document decision", "create DEC", "track assumption", "@rationale", "why did we choose"
Auto-detect and load minimal context from Skills folders using progressive disclosure, trimming 1400 tokens from system prompt through metadata-first loading. Use when optimizing token usage, managing multiple skills, or preventing context bloat. Load YAML metadata first, then conditionally load full SKILL.md only for matched skills based on task triggers. Achieves 10% accuracy improvement via targeted context. Triggers on "optimize skills", "reduce tokens", "smart loading", "skill efficiency".
