MCP server for the Dianshu (典枢) data platform, providing tools for authentication, order management, API calls, dataset search, and user profile access
19 tools with Chinese descriptions translated to English. All tools have basic descriptions (10-100 chars), but most lack depth and LLM-optimization guidance. Input schemas present for most tools but sparse: many parameters lack type constraints (enums, ranges). No output schemas documented. Error handling minimal, tools return raw strings without recovery guidance. Parameter descriptions are terse (avg ~30 chars vs baseline 72 chars). No tool annotations (readOnlyHint/destructiveHint). Composition is reasonable (single responsibility per tool), but chaining support unclear, output fields not documented, so downstream tool calls require guessing.
调用已购买的数据 API
检查典枢登录状态
清除登录状态,切换账号
通过任务编码下载并解密数据文件
获取数据 API 的详细信息及参数列表
获取数据集详情
获取典枢首页推荐数据
获取微信扫码登录二维码(仅返回二维码,不阻塞)
No output schemas documented. Tools return raw strings; LLMs cannot infer response structure, forcing unstructured parsing and wasting tokens.
Parameter descriptions are terse (avg ~30 chars vs baseline 72 chars). Missing format constraints, ranges, and dependency hints. E.g., 'timeout' lacks units clarification; 'params' in call_api lacks structure guidance.
No tool annotations (readOnlyHint/destructiveHint/idempotentHint). Agents cannot distinguish safe reads from destructive writes (e.g., delete_cookies, download_order, set_token lack destructiveHint).
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 56 | <=2025-11-25 | v2 |
获取我的个人资料
获取我的钱包信息
列出已购买的可下载数据产品
获取我的数据集列表
查询典枢订单列表
列出已购买的典枢 API 产品
获取钱包交易明细
打开可见浏览器窗口进行登录(支持扫码+账号密码)
搜索典枢数据集
手动设置登录 token(浏览器登录后的备选方案)
等待扫码登录完成
Error handling returns raw strings without recovery guidance. No categorization (retryable vs user-fixable vs fatal). E.g., 'login failed' tells agent nothing about next steps.
No pagination result limits documented. list_* tools accept pageNo/pageSize but lack guidance on max results per page or total count in response.