FastMCP 2.0 implementation providing production-level MCP tools for news aggregation, trending topic analysis, RSS feed management, and AI-powered news intelligence. Supports stdio and HTTP transport modes.
TrendRadar has 20 tools with moderate naming consistency but significant gaps in schema documentation, parameter descriptions, and output specifications. Tool names follow verb-noun convention reasonably well (get_, search_, analyze_, export_, etc.), and most include descriptions, though many descriptions are generic or insufficient for LLM decision-making. Critical issue: input schemas are inferred from code snippets rather than explicitly registered in visible tool definitions. Parameter type information is present in the search_tools.py excerpt but absent or incomplete for most other tools. Output schemas are not documented. Error handling exists but lacks actionable recovery guidance. This is a typical C-grade community server: functional but not production-ready.
分析指定话题或关键词的情感倾向
导出指定日期范围的新闻数据为多种格式(JSON、CSV、Excel等)
获取当前配置(特定章节或全部)
获取指定关键词或关注词在指定时间范围内的频率统计
获取最新一批爬取的新闻数据,快速了解当前热点
获取最新的 RSS 订阅数据(支持多日查询)。RSS 数据与热榜新闻分开存储,按时间流展示,适合获取特定来源的最新内容。
获取指定日期的新闻数据,用于历史数据分析和对比
获取多个平台的新闻对比分析
Input schemas not explicitly visible in tool registration. Schema definitions are inferred from code snippets (search_tools.py) rather than appearing as explicit JSON Schema in tool registration code. Most tools lack visible parameter type definitions and constraints.
Output schemas not documented. No tool documentation specifies what fields the tool returns or their types. LLMs cannot plan downstream calls or extract specific data from results.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 58 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 45 | - | v1 |
获取 RSS 源状态信息。查看当前配置的 RSS 源及其数据统计信息。
获取当前热门实体(人物、机构、地点等)及相关新闻
获取热点话题统计
列出本地或远程存储中可查询的数据日期
读取并解析新闻文章的详细内容
【推荐优先调用】将自然语言日期表达式解析为标准日期范围。为什么需要这个工具?用户经常使用"本周"、"最近7天"等自然语言表达日期,但 AI 模型自己计算日期可能导致不一致的结果。此工具在服务器端使用精确的当前时间计算,确保所有AI 模型获得一致的日期范围。
统一新闻搜索工具 - 整合多种搜索模式,支持同时搜索热榜和RSS
搜索 RSS 数据。在 RSS 订阅数据中搜索包含指定关键词的文章。
发送测试通知到所有配置的通知渠道
将本地数据同步至AWS S3远程存储
执行系统健康检查,返回各个组件的状态
更新配置文件中指定的设置
Vague parameter descriptions. Many parameter descriptions lack specificity about format, range, or valid values. E.g., 'mode' parameter in get_trending_topics says 'time mode' but does not clearly state the constraint is enum {daily, current}. Descriptions do not provide enough context for LLM decision-making.
Missing parameter type constraints. Parameters like 'platforms', 'feeds', 'format' lack enum/constraint definitions. LLMs must guess valid values instead of selecting from explicit options.
Destructive operations lack confirmation steps. Tools like 'export_data', 'update_config', 'sync_to_s3', and 'send_notification' modify state but have no dry-run or confirmation mechanism. Agent errors could cause unintended side effects.
Limited error handling documentation. Tool descriptions do not explain what errors can occur or how to recover. No mention of retryable vs. permanent failures, or what the agent should do if a tool fails.
Tool names and descriptions conflate multiple responsibilities. 'search_news_unified' combines keyword, fuzzy, and entity search modes, could be three separate tools. 'get_platform_comparison' conflates retrieval with comparison logic.
Pagination not clearly documented. Tools like 'get_latest_news', 'get_latest_rss' accept a 'limit' parameter but do not mention offset/cursor for pagination or total count in results. Large result sets risk context window overflow.