A Model Context Protocol server for NGA forum (Chinese gaming community). Provides tools to crawl posts, list forum topics, query board structures, and search for boards.
The NGA MCP server provides 4 tools for forum scraping with Chinese-language descriptions. All tools have descriptions and input schemas are present with type definitions. However, descriptions are verbose and include implementation details rather than LLM-optimized guidance. Parameter descriptions exist but lack constraint details (ranges, enums, format specifications). Output schemas are not formally documented. Error handling is basic (returns {success: false, error: str(e)}). Tool naming follows verb_noun pattern correctly. The server is functional but lacks production-grade polish in schema documentation and error guidance.
爬取 NGA 帖子内容 Args: url: 帖子链接(形如 https://bbs.nga.cn/read.php?tid=XXXX) topk: 返回评论条数上限,默认50条,0表示获取所有评论 Returns: 包含帖子标题、正文和评论列表的字典数据
根据板块名称或类别名称返回板块信息 支持两种查询模式: 1. 板块名称查询:返回最接近的TopK板块(模糊匹配,按编辑距离排序) 2. 类别名称查询:返回该类别下的所有板块 Args: name: 板块名称或类别名称(支持模糊匹配) topk: 返回数量上限,默认3个(类别查询时返回该类别所有板块) Returns: 包含板块详细信息(包含板面与合集)的字典数据
获取NGA论坛板块分类结构 建议首先调用此方法了解论坛整体结构,然后再使用get_board_links查询具体板块信息。 Returns: 包含按类别组织的板块名称列表的字典数据
抓取板块/板面/合集列表页的帖子 Args: url: 列表页链接(如 https://bbs.nga.cn/thread.php?fid=459 或 stid=43954481) topk: 主题条数上限,默认30 Returns: 包含帖子标题、回复数、发布时间等信息的字典数据
Output schemas not documented in tool descriptions or source. LLMs cannot infer what fields are returned, preventing downstream tool chaining and field extraction planning.
Parameter 'topk' lacks constraint documentation (valid range, what happens with extreme values like topk=0 on crawl_post). Descriptions mention '0表示获取所有评论' but only in Chinese and only for crawl_post, not documented as schema constraint.
Error handling returns bare exception strings without recovery guidance. 'Invalid URL format' or 'Board not found' would be more helpful than raw exception messages. No error classification (retryable vs user-fixable vs fatal).
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 53 | <=2025-11-25 | v2 |
| 2026-03-09 | F | 38 | - | v1 |
Descriptions are verbose and include implementation examples ('形如 https://bbs.nga.cn/read.php?tid=XXXX', 'fid=459') which LLMs may reuse literally rather than adapt to actual context.
All descriptions are in Chinese only. No English descriptions provided, limiting accessibility to English-speaking LLM agents and developers.
get_board_links parameter 'name' supports both fuzzy matching and category queries but this dual-mode behavior is not clearly documented as mutually exclusive modes. LLMs may be unsure which mode applies.