Raw JSON is hard to inspect when it arrives as one long string. This skill validates input and returns readable formatting with a chosen indentation size.

Web content is often hard to collect in a clean format for AI workflows. This skill uses Firecrawl to scrape, crawl, map, search, and extract structured website data.

App data files can become hard to maintain when unrelated records are grouped together. This skill gives simple rules for modular Next.js data files.

NBA data work often requires finding the right endpoint, parameters, and response format. This skill guides Claude, Codex, and Claude Code through nba_api extraction, transformation, and analysis workflows.

Parquet pipelines can waste storage and memory when defaults are used. This skill reviews Rust Parquet reads and writes, then suggests compression, sizing, projection, and filtering improvements.

Large data lakes can become slow, expensive, and hard to govern without clear architecture choices. This skill guides partitioning, storage layout, schema evolution, table formats, retention, and cost-aware lakehouse design.

Slow DataFusion queries can waste compute and hide simple plan issues. This skill reviews SQL and DataFrame patterns, then suggests concrete performance improvements.

Video conversion and transcription often require several command-line tools. This skill provides one CLI workflow for extracting audio, converting formats, and creating Whisper transcripts.

ML projects can look accurate while data leakage or weak evaluation hides serious flaws. This skill guides structured checks for splits, metrics, training, and deployment readiness.

Custom charts often need interactions that simple chart libraries cannot express. This skill guides D3.js SVG setup, data binding, scales, axes, tooltips, zoom, transitions, and color choices.

Data models and state updates can drift without clear review. This skill guides analysis of typing, mutation, and serialization choices.

Distributed neural network training requires coordinated setup, validation, and deployment. This skill guides Flow Nexus workflows from architecture design to production serving.

Agents often repeat weak strategies because learning data stays unstructured. This skill organizes trajectories, patterns, benchmarks, and recommendations into a ReasoningBank workflow.

ML model projects need a repeatable path from data preparation to deployment. This skill gives Claude, Codex, and Claude Code a structured workflow for training, evaluation, and packaging.

ML training can fail through NaN loss, overfitting, slow convergence, or unstable gradients. This skill guides diagnosis, targeted fixes, validation, and reporting.

Training failures are hard to debug when logs, curves, code, and configs point in different directions. This skill organizes the evidence and returns likely root causes, fixes, and verification steps.

Neural training across sandboxes is hard to coordinate. This skill gives Claude, Codex, and Claude Code workflows for Flow Nexus training, inference, templates, and clusters.

Large vector databases can become slow, memory-heavy, and hard to benchmark. This skill provides a structured AgentDB optimization workflow for quantization, HNSW indexing, caching, and validation.

Teams need fast semantic retrieval for documents, RAG systems, and knowledge bases. This skill gives AgentDB setup, search patterns, and tuning guidance.

Teams need relevant document retrieval for RAG and knowledge bases. This skill guides AgentDB setup, embedding storage, indexing, API search, and relevance tuning.