介绍
# Company Research
从公司 URL 生成专业排版 PDF 格式的综合账户调研报告。
## 工作流程
1. **调研**公司(网页抓取 + 搜索) 2. **构建** JSON 数据结构 3. 通过 `scripts/generate_report.py` **生成** PDF 4. 向用户 **交付** PDF
## 阶段 1:调研(并行)
并发执行这些搜索以最大限度地减少上下文使用:
``` WebFetch: [company URL] WebSearch: "[company name] funding news 2024" WebSearch: "[company name] competitors market" WebSearch: "[company name] CEO founder leadership" ```
从网站提取:公司名称、行业、总部、成立时间、领导层、产品/服务、定价模式、目标客户、案例研究、客户评价、近期新闻。
## 阶段 2:构建数据结构
创建符合此架构的 JSON(完整规范请参阅 `references/data-schema.md`):
```json { "company_name": "...", "source_url": "...", "report_date": "January 20, 2026", "executive_summary": "3-5 sentences...", "profile": { "name": "...", "industry": "...", ... }, "products": { "offerings": [...], "differentiators": [...] }, "target_market": { "segments": "...", "verticals": [...] }, "use_cases": [{ "title": "...", "description": "..." }], "competitors": [{ "name": "...", "strengths": "...", "differentiation": "..." }], "industry": { "trends": [...], "opportunities": [...], "challenges": [...] }, "developments": [{ "date": "...", "title": "...", "description": "..." }], "lead_gen": { "keywords": {...}, "outreach_angles": [...] }, "info_gaps": ["..."] } ```
## 阶段 3:生成 PDF
```bash # Install if needed pip install reportlab
# Save JSON to temp file cat > /tmp/research_data.json << 'EOF' {...your JSON data...} EOF
# Generate PDF python3 scripts/generate_report.py /tmp/research_data.json /path/to/output/report.pdf ```
## 阶段 4:交付
将 PDF 保存到工作区文件夹并提供下载链接: ``` [Download Company Research Report](computer:///sessions/.../report.pdf) ```
## 质量标准
- **准确性**:声明以可观察的证据为基础;引用来源 - **具体性**:包括产品名称、指标、客户示例 - **完整性**:将信息缺失注明为“非公开信息” - **无捏造**:绝不凭空捏造信息
## 资源
- `scripts/generate_report.py` - PDF 生成器(使用 reportlab) - `references/data-schema.md` - 带有示例的完整 JSON 架构