菜单
Training-free prompting strategy that pre-organizes query information into compact structured format, reducing generation tokens by 2-3× while maintaining reasoning accuracy. Apply when reasoning performance is bottlenecked by verbose input formatting.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/ADu2021/skillXiv --skill focused-chain-of-thought命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Convert arXiv papers into ready-to-use agent skills using category-aware extraction. First classifies the paper into one or more of 11 research categories, then applies a specialized extraction pipeline for each category — because different types of papers produce different types of usable knowledge. A single paper can yield multiple skills if it spans categories. Use this skill whenever the user wants to turn a paper into a skill, extract practical techniques from research, build a skill library from papers, convert arXiv papers into reusable agent instructions, or batch-process multiple papers into skills. Also trigger when someone asks about extracting actionable knowledge from papers, making research practical for LLM agents, or systematically converting academic contributions into structured agent capabilities.
Establish regret bounds for behavior cloning with discretized actions combining statistical error and quantization error terms. Prove smoothness requirements for safe quantizer design, show that learning-based quantizers fail these requirements, and propose model-based augmentation to reduce error dependence from H² to H.
Dynamically allocate LoRA ranks per-layer during fine-tuning instead of using fixed uniform ranks. Learn optimal rank for each layer and subject via variational framework with discretized exponential distribution, reducing memory footprint while maintaining fidelity and text-alignment.
基于 SOC 职业分类
正在显示 SKILL.md
| name | focused-chain-of-thought |
| title | Focused Chain-of-Thought: Structured Information Extraction for Efficient Reasoning |
| version | 0.0.2 |
| engine | skillxiv-v0.0.2-claude-opus-4.6 |
| license | MIT |
| url | https://arxiv.org/abs/2511.22176 |
| keywords | ["prompt-engineering","reasoning-efficiency","information-extraction","structured-formatting","token-reduction"] |
| description | Training-free prompting strategy that pre-organizes query information into compact structured format, reducing generation tokens by 2-3× while maintaining reasoning accuracy. Apply when reasoning performance is bottlenecked by verbose input formatting. |
Focused Chain-of-Thought (F-CoT) is a training-free prompting strategy that improves inference efficiency by separating information extraction from reasoning. The method first organizes relevant information from a query into a structured, compact format, then guides the model to reason exclusively over this organized context rather than the original verbose question.
The approach operates in two stages:
Stage 1 - Information Organization: Parse the original query and extract relevant facts, constraints, and context. Format these into a clean, structured representation (tables, lists, key-value pairs). This extraction removes verbal noise and redundancy.
Stage 2 - Focused Reasoning: Present the organized information to the model alongside a query, asking it to reason exclusively from the structured context. The model no longer needs to parse or filter through verbose original text.
This two-stage separation enables the model to focus computation on reasoning rather than information comprehension.
Extraction template: Design a format for your domain. Examples:
Structured representation: Convert original text into tables, bullet points, or key-value pairs. Each element should be a complete, self-contained fact.
Focused prompt: Chain-of-thought prompt operating only on structured context:
Context: [structured information]
Task: [reasoning goal]
Reasoning: