ltk

Use when evaluating agent performance, building test frameworks, measuring quality, or asking about "agent evaluation", "LLM-as-judge", "agent testing", "quality metrics", "evaluation rubrics", "agent benchmarks"

Use when "codex", "use gpt", "gpt-5", "openai codex", "let openai", "full-auto", "autonomous code generation"

Use when compressing agent context, implementing conversation summarization, reducing token usage in long sessions, or asking about "context compression", "conversation history", "token optimization", "context limits", "summarization strategies"

Use when diagnosing agent failures, debugging lost-in-middle issues, understanding context poisoning, or asking about "context degradation", "lost in middle", "context poisoning", "attention patterns", "context clash", "agent performance drops"

Use when optimizing agent context, reducing token costs, implementing KV-cache optimization, or asking about "context optimization", "token reduction", "context limits", "observation masking", "context budgeting", "context partitioning"

Use when "CrewAI", "multi-agent systems", "agent orchestration", "AI crews", or asking about "autonomous agents", "agent collaboration", "role-based agents", "agent workflows", "AI team coordination"

Use when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies

Use when "DSPy", "declarative prompting", "automatic prompt optimization", "Stanford NLP", or asking about "optimizing prompts", "prompt compilation", "modular LLM programming", "chain of thought", "few-shot learning"

Use when you have a written implementation plan to execute in a separate session with review checkpoints

Use when "LangChain", "LLM chains", "ReAct agents", "tool calling", or asking about "RAG pipelines", "conversation memory", "document QA", "agent tools", "LangSmith"

Use when building "MCP server", "Model Context Protocol", creating "Claude tools", "MCP tools", or asking about "FastMCP", "MCP SDK", "tool development for LLMs", "external API integration for Claude"

Use when implementing agent memory, persisting state across sessions, building knowledge graphs, tracking entities, or asking about "agent memory", "knowledge graph", "entity memory", "vector stores", "temporal knowledge", "cross-session persistence"

Use when designing multi-agent systems, implementing supervisor patterns, coordinating multiple agents, or asking about "multi-agent", "supervisor pattern", "swarm", "agent handoffs", "orchestration", "parallel agents"

Use when "nanobanana", "generate image", "create image", "edit image", "AI drawing", "Gemini image", "image generation"

This skill should be used when starting complex multi-step tasks, research projects, or any task requiring >5 tool calls. Implements Manus-style file-based planning with task_plan.md, findings.md, and progress.md.

Use when creating Claude Code plugins, writing skills, building commands, developing agents, or asking about "plugin development", "create skill", "write command", "build agent", "SKILL.md", "plugin structure", "progressive disclosure"

Use when "writing prompts", "prompt optimization", "few-shot learning", "chain of thought", or asking about "RAG systems", "agent workflows", "LLM integration", "prompt templates"

Use when executing implementation plans with independent tasks in the current session

Use when running /ltk:sync-submodules, updating submodules, or needing to "sync", "merge", "adapt", "learn from" other Claude Code plugins or repos

Use when designing agent tools, creating tool descriptions, implementing MCP tools, or asking about "tool design", "agent tools", "tool descriptions", "MCP", "function calling", "tool consolidation"

Use when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions

Use when you have a spec or requirements for a multi-step task, before touching code

Use when creating new skills, editing existing skills, or verifying skills work before deployment

Use when "Dask", "parallel computing", "distributed computing", "larger than memory", or asking about "parallel pandas", "parallel numpy", "out-of-core", "multi-file processing", "cluster computing", "lazy evaluation dataframe"

Use when "data pipelines", "ETL", "data warehousing", "data lakes", or asking about "Airflow", "Spark", "dbt", "Snowflake", "BigQuery", "data modeling"

Use when "statistical modeling", "A/B testing", "experiment design", "causal inference", "predictive modeling", or asking about "hypothesis testing", "feature engineering", "data analysis", "pandas", "scikit-learn"

Use when "experiment tracking", "MLflow", "Weights & Biases", "wandb", "model registry", "hyperparameter logging", "ML experiments", "training metrics"

Use when "GeoPandas", "geospatial", "GIS", "shapefile", "GeoJSON", or asking about "spatial analysis", "coordinate transformation", "spatial join", "choropleth map", "buffer analysis", "geographic data", "map visualization"

Use when "tokenizers", "HuggingFace tokenizer", "BPE", "WordPiece", or asking about "train tokenizer", "custom vocabulary", "tokenization", "subword", "fast tokenizer", "encode text"

Use when "LLM inference", "serving LLM", "vLLM", "llama.cpp", "GGUF", "text generation", "model serving", "inference optimization", "KV cache", "continuous batching", "speculative decoding", "local LLM", "CPU inference"

Use when "training LLM", "finetuning", "RLHF", "distributed training", "DeepSpeed", "Accelerate", "PyTorch Lightning", "Ray Train", "TRL", "Unsloth", "LoRA training", "flash attention", "gradient checkpointing"

Use when "deploying ML models", "MLOps", "model serving", "feature stores", "model monitoring", or asking about "PyTorch deployment", "TensorFlow production", "RAG systems", "LLM integration", "ML infrastructure"

Use when "Modal", "serverless GPU", "cloud GPU", "deploy ML model", or asking about "serverless containers", "GPU compute", "batch processing", "scheduled jobs", "autoscaling ML"

Use when "CLIP", "Whisper", "Stable Diffusion", "SDXL", "speech-to-text", "text-to-image", "image generation", "transcription", "zero-shot classification", "image-text similarity", "inpainting", "ControlNet"

Use when evaluating LLMs, running benchmarks like MMLU/HumanEval/GSM8K, setting up evaluation pipelines, or asking about "NeMo Evaluator", "LLM benchmarking", "model evaluation", "MMLU", "HumanEval", "GSM8K", "benchmark harnesses"

Use when "NetworkX", "graph analysis", "network analysis", "graph algorithms", "shortest path", "centrality", "PageRank", "community detection", "social network", "knowledge graph"

Use when "Polars", "fast dataframe", "lazy evaluation", "Arrow backend", or asking about "pandas alternative", "parallel dataframe", "large CSV processing", "ETL pipeline", "expression API"

Use when "PyMC", "Bayesian", "MCMC", "probabilistic programming", or asking about "Bayesian regression", "hierarchical model", "NUTS sampler", "posterior distribution", "prior predictive", "credible intervals", "uncertainty quantification"

Use when "RAG", "retrieval augmented generation", "LangChain", "LlamaIndex", "sentence transformers", "embeddings", "document QA", "chatbot with documents", "semantic search"

Use when "scientific computing", "astronomy", "astropy", "bioinformatics", "biopython", "symbolic math", "sympy", "statistics", "statsmodels", "scientific Python"