全部能力

Train Mixture of Experts (MoE) models using DeepSpeed or HuggingFace. Use when training large-scale models with limited compute (5× cost reduction vs dense…

Simple Preference Optimization for LLM alignment. Reference-free alternative to DPO with better performance (+6.4 points on AlpacaEval 2.0). No reference model…

Language-independent tokenizer treating text as raw Unicode. Supports BPE and Unigram algorithms. Fast (50k sentences/sec), lightweight (6MB memory),…

Fine-tune LLMs using reinforcement learning with TRL - SFT for instruction tuning, DPO for preference alignment, PPO/GRPO for reward optimization, and reward…

Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage…

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences…

Reserved and on-demand GPU cloud instances for ML training and inference. Use when you need dedicated GPU instances with simple SSH access, persistent…

Evaluates code generation models across HumanEval, MBPP, MultiPL-E, and 15+ benchmarks with pass@k metrics. Use when benchmarking code models, comparing coding…

Checks inline code documentation quality: WHY-not-WHAT, density, forbidden content, docstrings quality, actuality, legacy cleanup. Use when auditing comments…

Checks hierarchy, links, SSOT, compression, requirements compliance, freshness, legacy cleanup. Use when auditing documentation structure.

Checks document semantic content against SCOPE and project goals, coverage gaps, off-topic content, SSOT. Use when auditing documentation relevance.

Expert in graph database design and development with deep knowledge of graph modeling, traversals, query optimization, and relationship patterns. Specializes…

Verifies claims in .md files (paths, versions, counts, configs, endpoints) against codebase, cross-checks contradictions. Use when auditing docs accuracy.

Guides implementation of agent memory systems, compares production frameworks (Mem0, Zep/Graphiti, Letta, LangMem, Cognee), and designs persistence architectures for cross-session knowledge retention. Use when the user asks to "implement agent memory", "persist state across sessions", "build knowledge graph for agents", "track entities over time", "add long-term memory", "choose a memory framework", or mentions temporal knowledge graphs, vector stores, entity memory, adaptive memory, dynamic memory or memory benchmarks (LoCoMo, LongMemEval).

Use this skill when the user asks to "check data usage", "list TCO policies", "reduce Coralogix costs", "optimize observability spend", "lower our logging bill", "data budget exceeded", "TCO policy", "retention tier", "archive storage", "ingestion costs", "frequent search vs archive", "why is our bill so high", "spending too much on logs", "data retention settings", "cost analysis", "usage breakdown", "optimize log volume", "control data ingestion", "archive cold data", "billing units", "plan consumption", "daily plan", "overage", "PAYG", "usage anomaly", "usage trend", "cx_data_usage_units", or wants to investigate, analyze, or reduce Coralogix data costs.

Turn validated benchmark research into campaign briefs and concept candidates for short-form video production. Use this when you already have research…

Aggregate and display system metrics with anomaly detection for a time period

Upgrades npm/yarn/pnpm dependencies with breaking change handling. Use when updating JavaScript/TypeScript dependencies.

Upgrades Python pip/poetry/pipenv dependencies with breaking change handling. Use when updating Python dependencies.

Upgrades .NET NuGet packages with breaking change handling. Use when updating .NET dependencies.