Skip to main content

long-form-breadth-depth-coherence

Use this skill when the user wants a long report, article, monograph, or multi-thousand-word draft that still needs to stay organized and substantive. Trigger it for requests like “write a 4,000-word report,” “draft a 10,000-word article,” or “expand this into a full-length overview without repeating yourself.” It is especially useful when long outputs tend to become shallow, repetitive, or incoherent.

설치로 이동

소스 정보

저장소
Dingxingdi/paper_fast_search_backup
최근 소스 활동
2026년 4월 10일 01:27
감지된 SKILL.md 언어
영어
스타
0
포크
0

설치 방법

기본적으로 소스를 먼저 확인하는 Prompt가 선택됩니다. 직접 명령으로 전환하거나 로컬 사본을 다운로드할 수도 있습니다.

소스 파일 검토

설치 여부를 결정하기 전에 SKILL.md와 SkillsMP에 표시된 보조 파일을 읽어 보세요.

파일 탐색기
4 개 파일

SKILL.md 표시 중

SKILL.md
소스 지침 · 읽기 전용 미리보기
name
long-form-breadth-depth-coherence
description
Use this skill when the user wants a long report, article, monograph, or multi-thousand-word draft that still needs to stay organized and substantive. Trigger it for requests like “write a 4,000-word report,” “draft a 10,000-word article,” or “expand this into a full-length overview without repeating yourself.” It is especially useful when long outputs tend to become shallow, repetitive, or incoherent.
# Skill: long-form-breadth-depth-coherence ## 1. Capability Definition & Real Case * **Professional Definition**: The capability to generate long-form written documents that satisfy large output requirements while preserving global relevance, topic breadth, topic depth, local clarity, and cross-section coherence over extended length. * **Dimension Hierarchy**: Grounded Expository Writing->Long-form Expansion->long-form-breadth-depth-coherence ### Real Case **[Case 1]** * **Initial Environment**: The agent receives a blank long-form writing environment and a single historical request whose target length far exceeds ordinary one-pass responses. * **Real Question**: Write a 10,000-word article on the history of the Roman Empire. * **Real Trajectory**: The agent decomposes the task into a writing plan, assigns paragraph-level or section-level subtasks, writes serially with previous sections in context, and checks breadth, depth, and coherence while length accumulates. * **Real Answer**: A long historical article that approaches the requested size and remains organized across introduction, rise, institutions, expansion, crisis, decline, and legacy. * **Why this demonstrates the capability**: Reaching 10,000 words is not enough by itself. The article must also remain coherent, relevant, and non-repetitive across many sections. This case therefore probes the combined management of output length and document-level writing quality. --- **[Case 2]** * **Initial Environment**: The agent receives a long-form instruction to produce an educational or analytical report that explicitly requires more than 4,000 words and expects section-level continuity. * **Real Question**: Expand this short brief on urban heat mitigation into a full-length, 4,500-word public-facing report with clear sections and practical recommendations. * **Real Trajectory**: The agent creates section budgets, serially writes each section while retaining the prior context, introduces transitions between technical explanation and policy guidance, and performs anti-repetition checks before concluding. * **Real Answer**: A sectioned public-facing report that covers causes, measurement, health effects, infrastructure responses, policy tradeoffs, and implementation guidance. * **Why this demonstrates the capability**: The challenge is to sustain both breadth and depth over a long output without letting sections drift apart or echo one another. The report must not only be long but also structurally balanced and readable. This makes it a direct test of document-scale writing control. ## Pipeline Execution Instructions To synthesize data for this capability, you must strictly follow a 3-phase pipeline. **Do not hallucinate steps.** Read the corresponding reference file for each phase sequentially: 1. **Phase 1: Environment Exploration** Read the exploration guidelines to discover raw knowledge seeds: `references/EXPLORATION.md` 2. **Phase 2: Trajectory Selection** Once Phase 1 is complete, read the selection criteria to evaluate the trajectory: `references/SELECTION.md` 3. **Phase 3: Data Synthesis** Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data: `references/SYNTHESIS.md`
GitHub에서 보기