scientify 1.1.5 → 1.2.0

package/openclaw.plugin.json

@@ -6,7 +6,7 @@
   "skills": [
     "skills/idea-generation",
     "skills/research-pipeline",
-    "skills/literature-review",
-    "skills/arxiv"
+    "skills/literature-survey",
+    "skills/write-review-paper"
   ]
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "scientify",
-  "version": "1.1.5",
+  "version": "1.2.0",
   "description": "Scientify - AI-powered research workflow automation for OpenClaw. Includes idea generation, literature review, research pipeline skills, and arxiv tool.",
   "type": "module",
   "main": "dist/index.js",
@@ -32,13 +32,16 @@
   "scripts": {
     "build": "tsc",
     "clean": "rm -rf dist",
+    "dev": "tsc --watch",
+    "dev:reload": "nodemon --watch src --watch skills --ext ts,md,json --exec \"npm run build && pkill -USR1 -f 'openclaw.mjs gateway' || true\" --delay 500ms",
     "prepublishOnly": "npm run clean && npm run build"
   },
   "devDependencies": {
     "@sinclair/typebox": "^0.32.0",
+    "@types/node": "^20.0.0",
+    "nodemon": "^3.1.11",
     "tar": "^7.0.0",
-    "typescript": "^5.3.0",
-    "@types/node": "^20.0.0"
+    "typescript": "^5.3.0"
   },
   "peerDependencies": {
     "openclaw": ">=2024.0.0"

package/skills/literature-survey/SKILL.md

@@ -0,0 +1,285 @@
+---
+name: literature-survey
+description: "Comprehensive literature survey for a research domain. Automatically searches, filters, clusters, and iterates to discover relevant papers. Use for: exploring a new research area, collecting papers for a topic, building a literature database."
+metadata:
+  {
+    "openclaw":
+      {
+        "emoji": "🔍",
+      },
+  }
+---
+
+# Literature Survey
+
+Comprehensive literature discovery workflow for a research domain. This skill searches broadly, filters by relevance, clusters by direction, and iterates to ensure complete coverage.
+
+**Use this skill when:**
+- Exploring a new research area
+- Collecting all relevant papers for a topic
+- Building a comprehensive literature database
+- Need to understand the landscape of a field
+
+**NOT for:**
+- Summarizing papers you already have (use `/literature-review`)
+- Finding a specific paper (use `arxiv_search` directly)
+- Generating research ideas (use `/idea-generation`)
+
+---
+
+## Architecture: Isolated Sub-agent
+
+This survey runs in an **isolated sub-session** to avoid context pollution. The main session only receives the final report.
+
+```
+Main Session
+    ↓
+sessions_spawn(task: "执行文献调研...", label: "literature-survey")
+    ↓
+Sub-agent Session (隔离上下文)
+    ├── Phase 1: 生成检索词
+    ├── Phase 2: 批量检索
+    ├── Phase 3: 相关性筛选
+    ├── Phase 4: 聚类分组
+    ├── Phase 5: 迭代发现
+    └── Phase 6: 生成报告
+    ↓
+返回主 Session: 摘要 + 文件路径
+```
+
+---
+
+## When User Requests Literature Survey
+
+**Step 1: Spawn isolated sub-agent**
+
+When user says things like:
+- "调研 [topic] 领域的文献"
+- "帮我收集 [topic] 相关的论文"
+- "Survey papers on [topic]"
+
+Use `sessions_spawn` to run the survey in isolation:
+
+```
+sessions_spawn({
+  task: `你是一个文献调研专家。请为研究主题 "{TOPIC}" 执行完整的文献调研。
+
+## 调研目标
+{USER_REQUIREMENTS}
+
+## 执行流程
+
+### Phase 1: 生成检索词
+基于研究主题，生成 8-15 个检索词组合，覆盖：
+- 核心概念的不同表述
+- 相关技术方法
+- 应用场景
+- 英文和中文关键词（如适用）
+
+将检索词保存到 $WORKSPACE/survey/search_terms.json
+
+### Phase 2: 批量检索
+对每个检索词使用 arxiv_search tool：
+- max_results: 30-50 per query
+- 合并去重（按 arxiv_id）
+- 记录每篇论文的来源检索词
+
+将原始结果保存到 $WORKSPACE/survey/raw_results.json
+
+### Phase 3: 相关性筛选
+阅读每篇论文的标题和摘要，判断与 "{TOPIC}" 的相关性：
+- 5分：高度相关，直接研究此主题
+- 4分：相关，涉及关键方法或应用
+- 3分：部分相关，可作为参考
+- 2分：边缘相关
+- 1分：不相关
+
+保留 score >= 4 的论文。
+将筛选结果保存到 $WORKSPACE/survey/filtered_papers.json
+
+### Phase 4: 聚类分组
+分析筛选后论文的摘要，识别 3-6 个研究方向/子主题。
+为每个方向创建子文件夹并分配论文：
+
+$WORKSPACE/papers/
+├── {direction-1}/
+│   ├── paper_list.md
+│   └── [arxiv_ids...]
+├── {direction-2}/
+│   └── ...
+└── uncategorized/
+
+将聚类结果保存到 $WORKSPACE/survey/clusters.json
+
+### Phase 5: 迭代发现（1-2轮）
+检查高分论文的摘要，识别：
+- 提到的新方法名称
+- 引用的重要工作
+- 新的关键词
+
+如果发现新方向，补充检索并重复 Phase 2-4。
+最多迭代 2 轮。
+
+### Phase 6: 生成报告
+创建 $WORKSPACE/survey/report.md：
+
+# 文献调研报告: {TOPIC}
+
+## 调研概要
+- 检索词数量: X
+- 初始检索: Y 篇
+- 筛选后: Z 篇
+- 研究方向: N 个
+
+## 研究方向分布
+
+### 方向1: [名称]
+- 论文数量: X
+- 代表性工作: [列表]
+- 主要特点: [描述]
+
+### 方向2: [名称]
+...
+
+## 高影响力论文 (Top 10)
+| 排名 | 标题 | 年份 | 相关度 | 方向 |
+|-----|------|-----|-------|-----|
+| 1   | ... | ... | 5     | ... |
+
+## 研究趋势
+[基于论文年份分布的观察]
+
+## 发现的新方向
+[迭代中发现的额外关键词和方向]
+
+## 建议阅读顺序
+1. [入门级论文]
+2. [核心方法论文]
+3. [最新进展]
+
+---
+
+完成后，向主 session 报告：
+- 总共发现的论文数量
+- 识别的研究方向
+- 报告文件位置`,
+  label: "literature-survey-{TOPIC_SLUG}",
+  runTimeoutSeconds: 900,
+  cleanup: "keep"
+})
+```
+
+**Step 2: Wait and relay results**
+
+Sub-agent 完成后会自动 announce 结果到主 session。
+将结果摘要展示给用户，包括：
+- 发现的论文数量
+- 主要研究方向
+- 报告文件位置
+
+---
+
+## Workspace Structure
+
+```
+~/.openclaw/workspace/projects/{project-id}/
+├── project.json
+├── survey/                    # 调研过程数据
+│   ├── search_terms.json      # 检索词列表
+│   ├── raw_results.json       # 原始检索结果
+│   ├── filtered_papers.json   # 筛选后的论文
+│   ├── clusters.json          # 聚类结果
+│   ├── iterations.log         # 迭代记录
+│   └── report.md              # 最终报告
+├── papers/                    # 按方向组织的论文
+│   ├── {direction-1}/
+│   │   ├── paper_list.md
+│   │   └── 2401.12345/       # .tex 源文件
+│   ├── {direction-2}/
+│   └── uncategorized/
+└── ideas/                     # 后续 idea-generation 输出
+```
+
+---
+
+## Data Schemas
+
+### search_terms.json
+```json
+{
+  "topic": "battery life prediction",
+  "generated_at": "2024-01-15T10:00:00Z",
+  "terms": [
+    {"term": "battery remaining useful life", "category": "core"},
+    {"term": "lithium-ion degradation prediction", "category": "method"},
+    {"term": "SOH estimation neural network", "category": "technique"},
+    {"term": "EV battery health monitoring", "category": "application"}
+  ]
+}
+```
+
+### filtered_papers.json
+```json
+{
+  "filtered_at": "2024-01-15T10:30:00Z",
+  "total_raw": 245,
+  "total_filtered": 42,
+  "papers": [
+    {
+      "arxiv_id": "2401.12345",
+      "title": "...",
+      "abstract": "...",
+      "authors": ["..."],
+      "published": "2024-01-15",
+      "relevance_score": 5,
+      "source_terms": ["battery RUL", "degradation prediction"],
+      "notes": "直接研究锂电池RUL预测"
+    }
+  ]
+}
+```
+
+### clusters.json
+```json
+{
+  "clustered_at": "2024-01-15T11:00:00Z",
+  "clusters": [
+    {
+      "id": "data-driven",
+      "name": "数据驱动方法",
+      "description": "使用机器学习/深度学习的方法",
+      "paper_count": 15,
+      "paper_ids": ["2401.12345", "2401.12346", "..."],
+      "keywords": ["LSTM", "CNN", "transformer", "neural network"]
+    },
+    {
+      "id": "physics-based",
+      "name": "物理模型方法",
+      "description": "基于电化学机理的方法",
+      "paper_count": 8,
+      "paper_ids": ["..."]
+    }
+  ]
+}
+```
+
+---
+
+## Quick Mode (Without Sub-agent)
+
+For smaller surveys (< 50 papers), can run directly without spawning:
+
+User: "快速调研一下 [topic]，不超过 30 篇"
+
+→ Run Phase 1-4 directly in main session
+→ Skip iteration
+→ Generate simplified report
+
+---
+
+## Commands
+
+- "调研 [topic] 领域" → Full survey with sub-agent
+- "快速调研 [topic]" → Quick mode, 30 papers max
+- "继续上次的调研" → Resume from existing survey data
+- "扩展调研 [new direction]" → Add new search terms and iterate

package/skills/write-review-paper/SKILL.md

@@ -0,0 +1,391 @@
+---
+name: write-review-paper
+description: "Write a literature review or survey paper from EXISTING papers. Guides through reading, note-taking, synthesis, and structured writing. Use after /literature-survey has collected papers."
+metadata:
+  {
+    "openclaw":
+      {
+        "emoji": "📝",
+      },
+  }
+---
+
+# Literature Review Writing
+
+Guide for writing a structured literature review or survey paper from papers you've already collected. This skill helps with reading strategy, note organization, and academic writing.
+
+**Use this skill when:**
+- You have collected papers (via `/literature-survey` or manually)
+- Need to write a literature review section for a thesis/paper
+- Writing a standalone survey paper
+- Synthesizing findings across multiple papers
+
+**NOT for:**
+- Searching and discovering new papers (use `/literature-survey`)
+- Generating research ideas (use `/idea-generation`)
+
+---
+
+## Prerequisites
+
+Before starting, ensure you have:
+1. Papers collected in `$WORKSPACE/papers/`
+2. Ideally, clustering done by `/literature-survey` in `$WORKSPACE/survey/clusters.json`
+
+Check active project:
+```bash
+cat ~/.openclaw/workspace/projects/.active 2>/dev/null
+ls $WORKSPACE/papers/
+```
+
+---
+
+## Phase 1: Reading Strategy
+
+### 1.1 Triage Papers by Priority
+
+Based on clusters from survey, prioritize reading:
+
+| Priority | Criteria | Reading Depth |
+|----------|----------|---------------|
+| P1 (必读) | 高引用、奠基性工作、你的直接相关方向 | 精读全文 |
+| P2 (重要) | 主要方法论、重要实验结果 | 读摘要+方法+实验 |
+| P3 (参考) | 辅助材料、边缘相关 | 仅读摘要 |
+
+Create `$WORKSPACE/review/reading_plan.md`:
+
+```markdown
+# Reading Plan
+
+## P1 - 必读 (精读)
+- [ ] [paper_id]: [title] - [reason]
+- [ ] ...
+
+## P2 - 重要 (选读)
+- [ ] ...
+
+## P3 - 参考 (略读)
+- [ ] ...
+```
+
+### 1.2 Reading Notes Template
+
+For each paper, create `$WORKSPACE/review/notes/{paper_id}.md`:
+
+```markdown
+# [Paper Title]
+
+**ID**: [arxiv_id / DOI]
+**Authors**: [author list]
+**Year**: [year]
+**Venue**: [conference/journal]
+**Priority**: P1/P2/P3
+
+## One-sentence Summary
+[用一句话概括这篇论文的核心贡献]
+
+## Problem & Motivation
+- 研究什么问题？
+- 为什么重要？
+- 现有方法的不足？
+
+## Method
+### Core Idea
+[核心创新点]
+
+### Technical Approach
+[关键技术细节]
+
+### Key Equations
+```latex
+[重要公式]
+```
+
+## Experiments
+- **Datasets**: [使用的数据集]
+- **Baselines**: [对比方法]
+- **Metrics**: [评价指标]
+- **Key Results**: [主要结论]
+
+## Strengths
+1. [优点1]
+2. [优点2]
+
+## Limitations
+1. [局限1]
+2. [局限2]
+
+## Connections
+- 与 [paper_x] 的关系：[描述]
+- 改进了 [method_y]：[如何改进]
+- 被 [paper_z] 引用/扩展：[描述]
+
+## Quotes for Citation
+> "[重要原文]" (Section X, Page Y)
+
+## My Comments
+[你的思考、疑问、可能的改进方向]
+```
+
+---
+
+## Phase 2: Synthesis & Organization
+
+### 2.1 Build Comparison Table
+
+Create `$WORKSPACE/review/comparison.md`:
+
+```markdown
+# Method Comparison
+
+| Paper | Year | Category | Key Innovation | Dataset | Metric | Result |
+|-------|------|----------|----------------|---------|--------|--------|
+| [A]   | 2023 | Data-driven | ... | ... | RMSE | 0.05 |
+| [B]   | 2022 | Hybrid | ... | ... | RMSE | 0.08 |
+```
+
+### 2.2 Timeline Analysis
+
+Create `$WORKSPACE/review/timeline.md`:
+
+```markdown
+# Research Timeline
+
+## 2018-2019: 早期探索
+- [Paper A]: 首次提出 X 方法
+- [Paper B]: 引入 Y 技术
+
+## 2020-2021: 方法成熟
+- [Paper C]: 提出 SOTA 方法
+- ...
+
+## 2022-2023: 新趋势
+- [Paper D]: 开始关注 Z 问题
+- ...
+
+## Key Milestones
+1. [Year]: [Event/Paper] - [Significance]
+```
+
+### 2.3 Taxonomy Design
+
+Create `$WORKSPACE/review/taxonomy.md`:
+
+```markdown
+# Taxonomy of Approaches
+
+## Dimension 1: Method Type
+├── Data-driven
+│   ├── Statistical (e.g., GPR, SVM)
+│   ├── Deep Learning
+│   │   ├── CNN-based
+│   │   ├── RNN/LSTM-based
+│   │   └── Transformer-based
+│   └── Hybrid
+└── Model-based
+    ├── Electrochemical
+    └── Equivalent Circuit
+
+## Dimension 2: Data Source
+├── Laboratory Data
+├── Real-world Driving Data
+└── Synthetic Data
+
+## Dimension 3: Prediction Horizon
+├── Short-term (< 100 cycles)
+├── Medium-term (100-500 cycles)
+└── Long-term (> 500 cycles)
+```
+
+---
+
+## Phase 3: Writing Structure
+
+### 3.1 Survey Paper Template
+
+Create `$WORKSPACE/review/draft.md`:
+
+```markdown
+# [Survey Title]: A Comprehensive Review
+
+## Abstract
+[Background - 1 sentence]
+[Problem - 1 sentence]
+[What this survey covers - 2 sentences]
+[Key findings - 2 sentences]
+[Conclusion - 1 sentence]
+
+## 1. Introduction
+
+### 1.1 Background and Motivation
+[为什么这个领域重要？]
+[当前研究的热度和趋势]
+
+### 1.2 Scope of This Survey
+[本综述覆盖的范围]
+[与已有综述的区别]
+
+### 1.3 Contributions
+本文的主要贡献：
+1. [贡献1]
+2. [贡献2]
+3. [贡献3]
+
+### 1.4 Organization
+本文结构如下：
+- Section 2: [内容]
+- Section 3: [内容]
+- ...
+
+## 2. Background and Preliminaries
+
+### 2.1 Problem Definition
+[正式定义研究问题]
+
+### 2.2 Key Concepts
+[核心概念解释]
+
+### 2.3 Evaluation Metrics
+[常用评价指标]
+
+## 3. Taxonomy of Methods
+
+### 3.1 Category A: [Name]
+
+#### 3.1.1 Subcategory A.1
+[方法描述]
+[代表性工作]
+
+#### 3.1.2 Subcategory A.2
+...
+
+### 3.2 Category B: [Name]
+...
+
+## 4. Comparative Analysis
+
+### 4.1 Quantitative Comparison
+[对比表格]
+[结果分析]
+
+### 4.2 Qualitative Comparison
+[方法特点对比]
+[适用场景分析]
+
+## 5. Datasets and Benchmarks
+
+### 5.1 Public Datasets
+| Dataset | Size | Source | Features |
+|---------|------|--------|----------|
+| ... | ... | ... | ... |
+
+### 5.2 Benchmark Protocols
+[常用的实验设置]
+
+## 6. Challenges and Future Directions
+
+### 6.1 Open Challenges
+1. **Challenge 1**: [描述]
+2. **Challenge 2**: [描述]
+
+### 6.2 Emerging Trends
+1. **Trend 1**: [描述]
+2. **Trend 2**: [描述]
+
+### 6.3 Recommended Research Directions
+1. [方向1]
+2. [方向2]
+
+## 7. Conclusion
+[总结主要发现]
+[对领域的展望]
+
+## References
+[BibTeX entries]
+```
+
+### 3.2 Thesis Literature Review Template
+
+For a thesis chapter, use this structure:
+
+```markdown
+# Chapter 2: Literature Review
+
+## 2.1 Introduction
+[本章目标和结构]
+
+## 2.2 [Topic Area 1]
+[相关工作综述]
+
+## 2.3 [Topic Area 2]
+[相关工作综述]
+
+## 2.4 Summary and Research Gaps
+[总结现有工作的不足]
+[引出你的研究问题]
+```
+
+---
+
+## Phase 4: Writing Tips
+
+### Citation Density Guidelines
+
+| Section | Citation Density |
+|---------|------------------|
+| Abstract | 0 citations |
+| Introduction | 10-20 citations |
+| Background | 5-10 citations |
+| Main Survey | 50-100+ citations |
+| Conclusion | 2-5 citations |
+
+### Transition Phrases
+
+**Introducing similar work:**
+- "Similarly, [Author] proposed..."
+- "Following this direction, ..."
+- "Building upon [X], [Author] extended..."
+
+**Introducing contrasting work:**
+- "In contrast, [Author] argued..."
+- "However, [Author] took a different approach..."
+- "Unlike previous methods, ..."
+
+**Summarizing:**
+- "In summary, existing methods can be categorized into..."
+- "The key insight from these works is..."
+
+### Common Mistakes to Avoid
+
+1. **列举式写作** - 不要只是 "A did X, B did Y, C did Z"
+2. **缺乏比较** - 要分析方法之间的关系和区别
+3. **时态混乱** - 描述方法用现在时，描述实验结果用过去时
+4. **过度引用** - 不是每句话都需要引用
+5. **遗漏重要工作** - 确保覆盖领域的奠基性工作
+
+---
+
+## Output Files
+
+```
+$WORKSPACE/review/
+├── reading_plan.md       # 阅读计划
+├── notes/                # 阅读笔记
+│   ├── {paper_id}.md
+│   └── ...
+├── comparison.md         # 对比表格
+├── timeline.md           # 时间线分析
+├── taxonomy.md           # 分类体系
+├── draft.md              # 综述草稿
+└── bibliography.bib      # 参考文献
+```
+
+---
+
+## Commands
+
+- "帮我写综述" → Full workflow from reading to writing
+- "生成阅读计划" → Create reading_plan.md
+- "对比这些论文" → Generate comparison.md
+- "写综述草稿" → Generate draft.md
+- "润色这一段" → Polish specific section

package/skills/arxiv/SKILL.md

@@ -1,138 +0,0 @@
----
-name: arxiv
-description: "Search arXiv.org for academic papers using the built-in arxiv tool. Use for literature search, finding related work, and downloading paper sources."
-metadata:
-  {
-    "openclaw":
-      {
-        "emoji": "📚",
-      },
-  }
----
-
-# ArXiv Search
-
-Use the built-in `arxiv` tool to search for academic papers on arXiv.org.
-
-## Basic Search
-
-```
-arxiv query:"graph neural network" max_results:10
-```
-
-## Parameters
-
-| Parameter | Type | Description |
-|-----------|------|-------------|
-| `query` | string | Search query (required) |
-| `max_results` | number | Max papers to return (default: 10, max: 50) |
-| `sort_by` | string | Sort by: "relevance", "lastUpdatedDate", "submittedDate" |
-| `date_from` | string | Filter papers after this date (YYYY-MM-DD) |
-| `download` | boolean | Download .tex source files (default: false) |
-| `output_dir` | string | Directory for downloads (default: ~/.openclaw/workspace/papers/) |
-
-## Examples
-
-Search for recent transformer papers:
-```
-arxiv query:"transformer attention mechanism" sort_by:"submittedDate" max_results:5
-```
-
-Search with date filter:
-```
-arxiv query:"diffusion models" date_from:"2024-01-01"
-```
-
-Search and download .tex sources:
-```
-arxiv query:"transformer attention" max_results:5 download:true
-```
-
-## Output
-
-Returns JSON with:
-- `query`: The search query
-- `total`: Number of results
-- `papers`: Array of paper objects with:
-  - `title`: Paper title
-  - `authors`: List of authors
-  - `abstract`: Paper abstract
-  - `arxiv_id`: ArXiv ID (e.g., "2401.12345")
-  - `pdf_url`: Direct PDF link
-  - `published`: Publication date
-  - `categories`: ArXiv categories (e.g., "cs.LG", "cs.AI")
-
-When `download: true`:
-- `downloads`: Array of download results with:
-  - `arxiv_id`: Paper ID
-  - `format`: "tex" or "pdf" (fallback)
-  - `files`: List of downloaded files
-  - `error`: Error message if download failed
-- `output_dir`: Directory where files were saved
-
----
-
-## Downloading Paper Source (.tex)
-
-**IMPORTANT**: Prefer downloading .tex source over PDF for better AI readability.
-
-### Download .tex Source (Recommended - Use Tool)
-
-The easiest way is to use the `arxiv` tool with `download: true`:
-
-```
-arxiv query:"your search" max_results:5 download:true output_dir:"~/.openclaw/workspace/papers"
-```
-
-This automatically:
-1. Downloads .tex source from `https://arxiv.org/src/{arxiv_id}`
-2. Extracts tar.gz archives
-3. Falls back to PDF if .tex unavailable
-
-### Manual Download (Bash)
-
-If you need to download specific papers manually:
-
-```bash
-mkdir -p ~/.openclaw/workspace/papers/{arxiv_id}
-cd ~/.openclaw/workspace/papers/{arxiv_id}
-curl -L "https://arxiv.org/src/{arxiv_id}" -o source.tar.gz
-tar -xzf source.tar.gz
-```
-
-### Why .tex over PDF?
-
-| Format | AI Readability | Formulas | Structure |
-|--------|---------------|----------|-----------|
-| **.tex** | Excellent | Full LaTeX | Preserved |
-| .pdf | Poor (needs OCR) | Lost/garbled | Lost |
-
-### Fallback to PDF
-
-If .tex source is unavailable (404 error), fall back to PDF:
-```bash
-curl -L "https://arxiv.org/pdf/{arxiv_id}.pdf" -o ~/.openclaw/workspace/papers/{arxiv_id}.pdf
-```
-
----
-
-## Workspace Integration
-
-**Project-based workspace**: When using with `idea-generation` or `research-pipeline` skills, papers are stored per-project:
-
-```
-~/.openclaw/workspace/projects/
-├── .active                   # Current project ID
-├── {project-id}/             # e.g., nlp-summarization, cv-segmentation
-│   ├── project.json          # Project metadata
-│   ├── papers/               # Downloaded papers for THIS project
-│   │   ├── 2401.12345/       # Extracted .tex source
-│   │   │   ├── main.tex
-│   │   │   └── ...
-│   │   └── 2402.67890.pdf    # PDF fallback
-│   └── ...
-```
-
-**When called from idea-generation**: Use `output_dir: "$WORKSPACE/papers"` where `$WORKSPACE` is the active project directory.
-
-**Standalone usage**: Default `output_dir` is `~/.openclaw/workspace/papers/` (flat structure).

package/skills/literature-review/SKILL.md

@@ -1,261 +0,0 @@
----
-name: literature-review
-description: "Generate reading notes and summaries from EXISTING papers (PDF/.tex files user already has). Use for: summarize papers, create reading notes, write literature review section. Does NOT search for new papers or generate research ideas."
-metadata:
-  {
-    "openclaw":
-      {
-        "emoji": "📖",
-      },
-  }
----
-
-# Literature Review
-
-Generate structured notes and synthesis documents from academic papers. Use this skill when the user wants to:
-- Summarize papers they've collected
-- Create reading notes for a research topic
-- Write a literature review section
-- Compare methods across multiple papers
-
-## Workspace Convention (Project-based)
-
-**IMPORTANT**: OpenClaw uses project-based workspaces. Each research topic has its own project directory.
-
-### Check Active Project First
-
-Before starting, check the active project:
-```bash
-cat ~/.openclaw/workspace/projects/.active 2>/dev/null
-```
-
-If a project is active, use `$WORKSPACE = ~/.openclaw/workspace/projects/{project_id}/`.
-
-If no active project, use the flat structure: `~/.openclaw/workspace/`.
-
-### Project-based Structure (Recommended)
-
-```
-~/.openclaw/workspace/projects/{project-id}/
-├── project.json              # Project metadata
-├── papers/                   # Downloaded PDFs/tex files
-│   ├── 2401.12345/
-│   │   └── main.tex
-│   └── ...
-└── literature/               # Generated outputs
-    ├── notes/                # Per-paper notes
-    │   ├── 2401.12345.md
-    │   └── ...
-    ├── synthesis.md          # Cross-paper synthesis
-    ├── bibliography.bib      # BibTeX entries
-    └── review_draft.md       # Optional: formatted review
-```
-
-### Flat Structure (Fallback)
-
-```
-~/.openclaw/workspace/
-├── papers/
-└── literature/
-    ├── notes/
-    ├── synthesis.md
-    └── ...
-```
-
-**File existence = step completion.** Skip steps whose output already exists.
-
-**In the steps below**, `$WORKSPACE` refers to the active project directory or `~/.openclaw/workspace/` if no project is active.
-
-## Step 1: Gather Papers
-
-Check what papers are available:
-
-1. **Check active project first**: `cat ~/.openclaw/workspace/projects/.active`
-2. **Look in project papers directory**: `ls -la $WORKSPACE/papers/`
-3. Check if user provided URLs or arXiv IDs
-
-If no papers found, ask user to provide:
-- ArXiv IDs (e.g., "2401.12345")
-- PDF URLs
-- Local file paths
-
-## Step 2: Read and Annotate Each Paper
-
-For each paper, create `$WORKSPACE/literature/notes/<paper_id>.md`:
-
-First, ensure the output directory exists:
-```bash
-mkdir -p $WORKSPACE/literature/notes
-```
-
-```markdown
-# [Paper Title]
-
-**ArXiv/DOI**: [id]
-**Authors**: [list]
-**Year**: [year]
-**Venue**: [conference/journal if known]
-
-## TL;DR
-[1-2 sentence summary of the main contribution]
-
-## Problem Statement
-[What problem does this paper address?]
-
-## Method
-[Key approach, algorithm, or framework]
-
-### Core Idea
-[The central insight or innovation]
-
-### Technical Details
-[Important formulas, architectures, or algorithms]
-
-```latex
-[Key equations if applicable]
-```
-
-## Experiments
-- **Datasets**: [list]
-- **Baselines**: [list]
-- **Main Results**: [key numbers]
-
-## Strengths
-- [strength 1]
-- [strength 2]
-
-## Weaknesses / Limitations
-- [limitation 1]
-- [limitation 2]
-
-## Relevance to My Research
-[How does this paper relate to the user's work? Leave blank if unknown]
-
-## Key Quotes
-> "[Important quote from the paper]" (Section X)
-
-## References to Follow
-- [Paper A]: [why interesting]
-- [Paper B]: [why interesting]
-```
-
-### Reading Strategy by Format
-
-| Format | Method |
-|--------|--------|
-| `.tex` | Use `read` directly. Search for `\section`, `\begin{equation}` |
-| `.pdf` | Use `read` (OpenClaw supports PDF). Focus on abstract, intro, method, experiments |
-| URL | Use `web_fetch` to get content, then summarize |
-
-### Quality Checklist
-
-Before finishing a note, verify:
-- [ ] TL;DR captures the main contribution
-- [ ] Method section explains the approach clearly
-- [ ] At least 2 strengths and 2 limitations identified
-- [ ] Key equations/algorithms included if applicable
-
-## Step 3: Generate BibTeX
-
-Create `$WORKSPACE/literature/bibliography.bib`:
-
-```bibtex
-@article{author2024title,
-  title={Full Paper Title},
-  author={Last, First and Last2, First2},
-  journal={arXiv preprint arXiv:2401.12345},
-  year={2024}
-}
-```
-
-For arXiv papers, use this format. For published papers, include venue, volume, pages.
-
-## Step 4: Synthesize Across Papers
-
-Create `$WORKSPACE/literature/synthesis.md`:
-
-```markdown
-# Literature Synthesis: [Topic]
-
-## Overview
-[Brief introduction to the research area]
-
-## Taxonomy of Approaches
-
-### Category A: [Name]
-Papers: [list]
-Key characteristics: [describe]
-
-### Category B: [Name]
-Papers: [list]
-Key characteristics: [describe]
-
-## Comparison Table
-
-| Paper | Method | Dataset | Key Metric | Result |
-|-------|--------|---------|------------|--------|
-| [A]   | ...    | ...     | ...        | ...    |
-| [B]   | ...    | ...     | ...        | ...    |
-
-## Evolution of Ideas
-[How has the field progressed? What are the trends?]
-
-## Open Problems
-1. [Gap 1]
-2. [Gap 2]
-
-## Recommendations
-[Which papers to read first? Which approaches are most promising?]
-```
-
-## Step 5 (Optional): Draft Literature Review
-
-If user requests a formal review, create `$WORKSPACE/literature/review_draft.md`:
-
-```markdown
-# Literature Review: [Topic]
-
-## 1. Introduction
-[Context and motivation for the review]
-
-## 2. Background
-[Essential concepts the reader needs]
-
-## 3. Survey of Methods
-
-### 3.1 [Category A]
-[Describe approaches in this category, cite papers]
-
-### 3.2 [Category B]
-[Describe approaches in this category, cite papers]
-
-## 4. Empirical Comparison
-[Summarize experimental findings across papers]
-
-## 5. Discussion
-[Trends, gaps, and future directions]
-
-## 6. Conclusion
-[Key takeaways]
-
-## References
-[BibTeX citations]
-```
-
-## Batch Processing
-
-If reviewing more than 10 papers:
-1. First pass: Generate TL;DR only for all papers
-2. User selects which papers need full notes
-3. Second pass: Full notes for selected papers
-4. Final pass: Synthesis
-
-Do NOT process all papers with full detail in a single session—context will overflow.
-
-## Commands
-
-User can say:
-- "Review these papers" → Full workflow
-- "Just summarize [paper]" → Single paper note
-- "Compare [paper A] and [paper B]" → Focused comparison
-- "Write a literature review on [topic]" → Full review draft