fiftyone-dataset-import
Universal dataset import for FiftyOne supporting all media types (images, videos, point clouds, 3D scenes), all label formats (COCO, YOLO, VOC, CVAT, KITTI, etc.), and multimodal grouped datasets. Use when users want to import any dataset regardless of format, automatically detect folder structure, handle autonomous driving data with multiple cameras and LiDAR, or create grouped datasets from multimodal data. Requires FiftyOne MCP server.
when-analyzing-performance-use-performance-analysis
Comprehensive performance analysis, bottleneck detection, and optimization recommendations for Claude Flow swarms
ai-runtime-toolkit
AI Runtime工具装备系统，支持8个内部专业工具和10+个外部CLI工具的整合管理，提供工具发现、执行和配置功能，遵循整合优于创造的设计理念
using-superpowers
Use when starting any conversation - establishes mandatory workflows for finding and using skills, including using Read tool before announcing usage, following brainstorming before coding, and creating TodoWrite todos for checklists
when-optimizing-agent-learning-use-reasoningbank-intelligence
Implement adaptive learning with ReasoningBank for pattern recognition, strategy optimization, and continuous improvement
prompt-engineering
Advanced prompt engineering techniques for optimal AI responses. Use this when crafting prompts, optimizing AI interactions, or designing system prompts for applications.
ml-training-debugger
Diagnose machine learning training failures including loss divergence, mode collapse, gradient issues, architecture problems, and optimization failures. This skill spawns a specialist ML debugging ...
harness-model-protocol
Analyze the protocol layer between agent harness and LLM model. Use when (1) understanding message wire formats and API contracts, (2) examining tool call encoding/decoding mechanisms, (3) evaluating streaming protocols and partial response handling, (4) identifying agentic chat primitives (system prompts, scratchpads, interrupts), (5) comparing multi-provider abstraction strategies, or (6) understanding how frameworks translate between native LLM APIs and internal representations.
context-optimization
Apply optimization techniques to extend effective context capacity. Use when context limits constrain agent performance, when optimizing for cost or latency, or when implementing long-running agent systems.
memory-system
Complete guide to the AgenticFleet memory system. Read this first.
prompt-architect
Comprehensive framework for analyzing, creating, and refining prompts for AI systems. Use when creating prompts for Claude, ChatGPT, or other language models, improving existing prompts, or applying evidence-based prompt engineering techniques. Applies structural optimization, self-consistency patterns, and anti-pattern detection to transform prompts into highly effective versions.
parallel-swarm-implementation
Loop 2 of the Three-Loop Integrated Development System. META-SKILL that dynamically compiles Loop 1 plans into agent+skill execution graphs. Queen Coordinator selects optimal agents from 86-agent registry and assigns skills (when available) or custom instructions. 9-step swarm with theater detection and reality validation. Receives plans from research-driven-planning, feeds to cicd-intelligent-recovery. Use for adaptive, theater-free implementation.
building-rag-systems
Build production RAG systems with semantic chunking, incremental indexing, and filtered retrieval. Use when implementing document ingestion pipelines, vector search with Qdrant, or context-aware retrieval. Covers chunking strategies, change detection, payload indexing, and context expansion. NOT when doing simple similarity search without production requirements.
extract-transcripts
Extract readable transcripts from Claude Code and Codex CLI session JSONL files
claude-opus-4-5-migration
將提示詞和程式碼從 Claude Sonnet 4.0、Sonnet 4.5 或 Opus 4.1 遷移至 Opus 4.5。當使用者想要更新其程式碼庫、提示詞或 API 呼叫以使用 Opus 4.5 時使用。處理模型字串更新和針對已知 Opus 4.5 行為差異的提示詞調整。不會遷移 Haiku 4.5。
when-analyzing-user-intent-use-intent-analyzer
Advanced intent interpretation system using cognitive science principles and probabilistic intent mapping
fiftyone-dataset-inference
Create a FiftyOne dataset from a directory of media files (images, videos, point clouds), optionally import labels in common formats (COCO, YOLO, VOC), run model inference, and store predictions. Use when users want to load local files into FiftyOne, apply ML models for detection, classification, or segmentation, or build end-to-end inference pipelines.
metacognitive-guard
Monitors Claude's responses for struggle signals and suggests escalation to deep-thinking agents when complexity exceeds comfortable reasoning capacity.
toon-format
TOON (Token-Oriented Object Notation) encoding for LLM-efficient data representation. 30-60% token savings vs JSON for structured data.
agent-creator
Creates specialized AI agents with optimized system prompts using the official 4-phase SOP methodology from Desktop .claude-flow, combined with evidence-based prompting techniques and Claude Agent SDK implementation. Use this skill when creating production-ready agents for specific domains, workflows, or tasks requiring consistent high-quality performance with deeply embedded domain knowledge.
creating-skills
Use when creating new Claude Code skills or improving existing ones - ensures skills are discoverable, scannable, and effective through proper structure, CSO optimization, and real examples
ml-antipattern-validator
Prevents 30+ critical AI/ML mistakes including data leakage, evaluation errors, training pitfalls, and deployment issues. Use when working with ML training, testing, model evaluation, or deployment.