219 skills · sorted by relevance
Measure code performance and run benchmarks to investigate slow operations, regressions, or memory pressure. Establish a baseline, compare results under matching conditions, report variability, and rerun correctness tests after optimization.
Tune AgentDB memory use and search speed through quantization, which stores vectors more compactly, search indexes, caching, and batch operations. Includes configuration examples and benchmarks to help balance accuracy, speed, and storage as datasets grow.
Improve slow Apache Spark jobs by reviewing data partitions, caching, memory settings, and shuffles, which move data between workers. Use the guidance to investigate bottlenecks and tune pipelines for larger datasets.
Fine-tune large language models with LoRA and QLoRA while limiting GPU memory use. Guides training settings and adapter management, using small sets of task-specific weights that can be saved, switched, or merged into a base model.
Get guidance on adapting LLMs with Unsloth, including implementation, API usage, and debugging. Covers LoRA and QLoRA, methods for adjusting models with fewer trainable parameters, with a focus on training speed and memory use.
Use past local Codex sessions to propose updates to a selected skill and project memory file. Recurring tasks are replayed, changes are checked on separate validation tasks, and proposals are staged for review before adoption.
Helps fine-tune large language models with LoRA and QLoRA, methods that train small sets of additional parameters. Covers memory-conscious training, configuration choices, saving adapters, and switching task-specific adapters on a shared base model.
Write and refactor C++20/23 applications, addressing memory management, concurrency, and performance bottlenecks. Produce interfaces, implementations, tests, and relevant CMake updates, then check runtime errors and measure performance before applying targeted optimizations.
Performance optimization specialist for profiling, caching, and latency optimizationUse when "performance, latency, slow query, profiling, caching, optimization, N+1, connection pool, p99, performance, profiling, caching, latency, optimization, async, database, load-testing, ml-memory" mentioned.
The English summary is being prepared. Showing the original description.
Keeping codebases healthy, performant, and maintainable - refactoring, performance optimization, and technical debt managementUse when "refactor, optimize, performance, technical debt, cleanup, architecture, speed up, bundle size, memory leak, slow query, code smell, complexity, dead code, performance, refactoring, optimization, technical-debt, architecture, cleanup, bundle, memory" mentioned.
The English summary is being prepared. Showing the original description.
Generates a structured Project Brief — a comprehensive markdown document that extracts goals, architecture, knowledge file inventory, Custom Instructions summary, Memory contents, current state, ecosystem position, and key decisions from a Claude Project. Briefs serve as uploadable context documents: add one to any other Project for immediate deep awareness of the source Project's purpose, architecture, and progress. Use when user says "create a brief," "brief this project," "extract project context," "generate a project summary for another project," "I need to share this project's context," "prepare this project for cross-project reference," or "document this project." Also use when the user is preparing to work across Projects and needs portable context. Do NOT use for session handoffs, project audits, or Memory optimization (use rootnode-session-handoff, rootnode-project-audit, or rootnode-memory-optimization respectively, if available).
The English summary is being prepared. Showing the original description.
Expert in optimization methods covering linear programming, convex optimization, gradient methods, and constrained optimization
The English summary is being prepared. Showing the original description.
Quantizes LLMs to 8-bit or 4-bit for 50-75% memory reduction with minimal accuracy loss. Use when GPU memory is limited, need to fit larger models, or want faster inference. Supports INT8, NF4, FP4 formats, QLoRA training, and 8-bit optimizers. Works with HuggingFace Transformers.
The English summary is being prepared. Showing the original description.
Performance optimization and measurement for .NET applications. Navigation skill covering Span, ArrayPool, memory management, benchmarking, profiling, Native AOT, and optimization patterns. For building high-performance applications. Keywords: performance, optimization, span, arraypool, benchmarking, profiling, memory, gc, aot, native-aot
The English summary is being prepared. Showing the original description.
Investigate performance bottlenecks in REA analysis workflows and improve caching, queries, or shared processing. Compare completed tasks before and after changes while preserving correct results, examined coverage, and access to supporting evidence.
Integrate Flash Attention into Transformer models to speed up attention, the calculation that relates input elements, while reducing GPU memory use. Covers PyTorch integration, performance measurement, output comparison, and a separate H100 optimization path.
Improve the MCP server that connects claude-flow v3 to tools. It guides connection reuse, faster tool lookup, request distribution, and communication changes, with monitoring to evaluate response times, errors, memory use, and cache performance.
Turn specifications into pseudocode, a language-independent outline of program logic. Define data structures, handle edge cases, analyze time and memory requirements, and document design patterns and optimization opportunities before implementation.
Improve the MCP server that connects claude-flow v3 to tools. It guides connection reuse, faster tool lookup, request distribution, and communication changes, with monitoring to evaluate response times, errors, memory use, and cache performance.
Measure Python execution time and memory use to locate slow or resource-heavy code. Use profiling results to guide improvements in data structures, caching, input and output operations, algorithms, and parallel processing.
Train language models with SimPO using pairs of preferred and rejected answers. Configure preference training without a separate reference model, adapt settings for base or instruction models, and address unstable training, memory limits, or lost capabilities.
Helps configure and troubleshoot DeepSpeed for training large models across multiple devices. Covers memory optimization, parallel training, and DeepNVMe storage transfers, with examples for tuning data movement and configuring models that divide work among specialized components.
Review code for performance bottlenecks across database queries, memory use, frontend rendering, network requests, and algorithms. Use a structured checklist to identify optimization opportunities and classify their severity according to expected load and scale.
Review game code for likely performance bottlenecks in processing, memory, rendering, and data loading. Compare available estimates with configured limits and produce ranked optimization recommendations, including expected impact, effort, risks, and areas needing runtime measurement.