idea-generator

name	idea-generator
description	Analyze connections across read papers and generate actionable research ideas. Use when user says: 给我一些研究灵感, idea generator, 跨论文分析, research ideas.

Idea Generator — Cross-Paper Insights

⚙️ Step 0 — Read the Research Profile (Always First)

Before running any capability, load the user's research profile.

Location: ~/.openclaw/workspace/research-claw-config.md

If this file does not exist, use these defaults silently and mention at the end:

💡 想定制推荐兴趣？试试说「更新我的研究画像」

# Default profile (used when no config found)
research_direction: "Large language models, reinforcement learning, agentic AI"
seed_papers: []
keywords:
  - large language models
  - reinforcement learning
  - agentic AI / AI agents
  - retrieval-augmented generation
  - multimodal models
whitelist_authors: []
learned_preferences:
  accept: []
  reject: []

Config fields reference:

• research_direction — free-text description of the user's research focus
• seed_papers — list of arXiv IDs the user considers gold-standard references
• keywords — interest topics used for Paper Scout search queries
• whitelist_authors — researcher names to prioritize in recommendations
• learned_preferences.accept — keywords/topics user has explicitly liked
• learned_preferences.reject — keywords/topics user has skipped or disliked

---

---

Goal: Analyze connections across the user's read papers and generate 3–5 actionable research ideas.

Triggers: 给我一些研究灵感 · idea generator · 跨论文分析 · research ideas

Step 1 — Gather paper corpus

1. Load ~/.openclaw/workspace/research-claw-reading-list.json
2. Filter papers with status: "done" or status: "reading" (i.e., papers the user has actually engaged with)
3. For each paper, load its HTML note from the note_link field (if available) or re-fetch abstract from arXiv
4. Extract: title, method summary, results, takeaway, tags for each paper

If fewer than 3 papers in the corpus, inform the user:

💡 你当前已读论文较少（{N} 篇）。建议先用"帮我读一下"深读几篇，效果会更好！以下基于现有内容生成。

Step 2 — Identify research gaps

Analyze the collected papers for:

• Recurring unsolved problems: limitations mentioned across multiple papers
• Methodology mismatches: technique A works in domain X but hasn't been tried in domain Y
• Evaluation gaps: common benchmarks that none of the papers tackle
• Combination opportunities: two methods from different papers that could be synergistically combined
• Scaling questions: results that hold at small scale but haven't been validated at large scale

Step 3 — Generate idea proposals

Produce 3–5 research ideas. For each idea, output:

💡 Research Idea {N}: **{CATCHY_TITLE}**

📋 **Problem Statement**
{1-2 sentences: what specific gap or challenge does this idea address?}

🔧 **Proposed Approach**
{2-3 sentences: how would you tackle it? What's the key insight?}

📚 **Related Papers**
• {Paper 1 title} — {how it connects}
• {Paper 2 title} — {how it connects}

🎯 **Feasibility Assessment**
- Dataset: {what data is needed / what's available}
- Compute: {estimated compute requirements}
- Timeline: {rough estimate for a first experiment}
- Risk: {main technical risk}

⭐ **Why Now?** {1 sentence — why this is timely / what makes it tractable today}

Step 4 — Output

Format all ideas in sequence. End with:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
💾 想把某个灵感发展成论文？说 "帮我写论文" 并描述你选中的想法！

---

---

⚠️ Error Handling

Error	Handling
arXiv API returns empty results	Retry once with broader query; if still empty, note "arXiv API temporarily unavailable"
PDF tool times out	Fall back to abstract-only mode; note [Abstract only — PDF timeout] in the note
PDF tool returns error for a paper	Try fetching https://ar5iv.labs.arxiv.org/html/{ARXIV_ID} as HTML fallback
Config file missing	Use defaults silently; add a note at end: "💡 想定制？说「更新我的研究画像」"
Reading list JSON missing or malformed	Start fresh with an empty list; inform user: "未找到现有列表，已新建空列表"
Template file not found	Report the expected path and ask user to check installation
No papers in last 3 days	Extend to 7 days, note it: "（近3天论文较少，已扩展至7天）"
Fewer than 3 read papers for Idea Generator	Proceed anyway, but note the limitation
User provides PDF/DOI instead of arXiv	Try to extract arXiv ID from DOI or search arXiv by title

---