Slow Python code wastes engineering time and can hide the real bottleneck. This skill guides Claude, Codex, and Claude Code through profiling, benchmarking, memory checks, and focused optimization patterns.

Teams need reliable structured data from unstructured text. This skill shows how to use Cohere v2, Python, and JSON Schema mode for validated extraction workflows.

Engineering agents can make broad or risky changes when project expectations are unclear. This skill keeps work focused on small patches, existing patterns, and clear validation.

Research reviews often mix evidence, opinion, and unsupported claims. This skill helps Claude, Codex, and Claude Code summarize provided sources with neutral wording and clear evidence boundaries.

Local application stacks can be hard to debug when restarts and config changes happen without context. This skill guides Claude, Codex, and Claude Code toward logs, status checks, clear impact notes, and reversible operations.

Complex evidence workflows can lose context, citations, and ownership across multiple agents. This skill creates a plan-first coordination layer for tests, delegation, audits, and quality control.

Large reports are difficult to edit without losing structure or citations. This skill guides chunked Markdown edits, backups, and citation-aware review.

Long documentation wastes context and slows AI-assisted work. This skill turns project knowledge into concise tables, bullets, examples, and agent-ready structure.

Repeated assistant mistakes slow down development and reduce trust. This skill turns each mistake into a compact local rule for future work.

Linux performance problems are hard to triage from raw monitoring data. This skill helps Claude, Codex, and Claude Code fetch or interpret GASP metrics and turn them into clear diagnostic findings.

Unclear requirements cause wasted generation cycles and rework. This skill checks inputs against saved standards before a coding or content pipeline continues.

Final reviews often miss gaps in correctness, security, and maintainability. This skill applies a six-dimension checklist and clear scoring before delivery.

Chapter validation often leaves results scattered across multiple files. This skill combines validation outputs into clear reports, metrics, and deployment recommendations.

PDF conversion loses layout context when text, images, and page numbers are handled separately. This skill creates persistent extraction artifacts that support layout-aware processing in Claude, Codex, and Claude Code.

Generated HTML can contain missing tags, weak document structure, or incomplete metadata. This skill checks page structure and produces a validation report before later pipeline steps run.

Teams often lose time turning approved requirements into complete work products. This skill guides Claude, Codex, or Claude Code to create code, tests, documentation, or content that follows saved standards.

Teams often receive useful output with uneven formatting, headings, spacing, or code style. This skill applies stated standards and returns a polished deliverable ready for review.

Commit messages often lose important context after code changes are staged. This skill reviews staged Git changes and drafts a clear message for human approval.

Writing useful commit messages from staged changes takes time and context switching. This skill collects Git status, statistics, and diff context so Claude, Codex, or Claude Code can draft a reviewable message.

Teams often write code that only developers can follow. This skill checks names, comments, jargon, and documentation so broader teams can understand code faster.