// Collect and research materials for X (Twitter) content creation using multi-round web search strategy. Use when user wants to gather trending topics, research subjects for X posts, or mentions "collect materials", "research topic", "find content for X", "x-collect". Performs 4-round deep research mimicking human research workflow.
Create viral X (Twitter) posts including short tweets, threads, and replies. Use when user wants to write X content, create posts, or mentions "create tweet", "write thread", "x-create", "写推文", "创作推文". Supports 5 post styles with customizable templates, plus a mandatory humanize pass to reduce AI-sounding phrasing. First-time users go through onboarding to set up profile.
Score and filter topics for X content creation using weighted criteria. Use when user wants to evaluate collected materials, filter topics by score, or mentions "filter topics", "score materials", "x-filter", "选题筛选". Applies 10-point scoring system with customizable weights.
Publish tweets and threads to X (Twitter) draft using browser automation. Use when user wants to publish content to X, save to drafts, or mentions "publish to X", "post tweet", "x-publish", "发布推文". Supports short tweets and threads. NEVER auto-publish, always saves to draft.
| name | x-collect |
| description | Collect and research materials for X (Twitter) content creation using multi-round web search strategy. Use when user wants to gather trending topics, research subjects for X posts, or mentions "collect materials", "research topic", "find content for X", "x-collect". Performs 4-round deep research mimicking human research workflow. |
Collect trending topics and research materials for X content creation using a systematic 4-round web search strategy.
If available, leverage user preference state to improve search quality:
~/.claude/skills/x-create/state/liked_topics.json (preferred topics/entities/sources)rejected_topics.json (avoid low-value/repetitive topics)If state files don't exist, proceed normally.
User provides:
Simulate human research thinking process with progressive depth.
Query expansion rules (lightweight):
topicDedup & clustering (at the end of each round):
Round 1: Official Sources (权威信息)
Search: "{topic} 官方文档"
Search: "{topic} GitHub"
Search: "{topic} official announcement"
Goal: Get authoritative first-hand information
Round 2: Technical Analysis (技术解析)
Search: "{topic} 详细介绍"
Search: "{topic} 教程 tutorial"
Search: "{topic} how it works"
Goal: Understand technical details and mechanisms
Round 3: Comparison & Reviews (对比评测)
Search: "{topic} vs {competitor}"
Search: "{topic} 评测 review"
Search: "{topic} pros cons"
Goal: Get different perspectives and comparisons
Round 4: Supplementary Verification (补充验证)
# Analyze gaps from previous rounds
missing_info = analyze_gaps(previous_results)
Search: "{missing_info}"
Search: "{topic} 最新 latest 2024 2025"
Goal: Fill information gaps and get latest updates
Generate structured material document:
# {Topic} 素材收集报告
## 收集时间
{timestamp}
## 核心信息
- **官方定义**: ...
- **关键特性**: ...
- **最新动态**: ...
## 素材列表
### 素材 1
- **标题**: ...
- **来源**: {url}
- **摘要**: 2-3句话概括
- **关键点**:
- 要点1
- 要点2
- **潜在选题角度**: ...
- **推荐推文类型**: [高价值干货/犀利观点/热点评论/故事洞察/技术解析]
- **QualityScore (0-10)**: ...
- **TopicCluster**: ...
- **HookCandidates**:
- ...
### 素材 2
...
## 热度分析
- **当前热度**: 高/中/低
- **趋势**: 上升/稳定/下降
- **讨论焦点**: ...
## 争议点
- 争议1: ...
- 争议2: ...
## 下一步建议
使用 `/x-filter` 对素材进行打分筛选
Additionally, append a machine-readable block for hooks/state ingestion:
MATERIALS_JSON
{
"schema_version": "x_skills.materials.v1",
"topic": "{Topic}",
"timestamp": "{timestamp}",
"items": [
{
"title": "...",
"url": "...",
"summary": "...",
"key_points": ["..."],
"quality_score_0_10": 0,
"topic_cluster": "...",
"hook_candidates": ["..."]
}
]
}
python ~/.claude/skills/x-create/scripts/x_state.py event --event collect.completed --payload-json '{"topic":"...","items_count":12,"explore_ratio":0.2}'User: /x-collect Claude MCP协议
Expected behavior:
After collection, suggest:
素材收集完成！共找到 X 条相关素材。
下一步：运行 /x-filter 对素材进行打分筛选，≥7分的选题将进入创作池。