A collection of MCP servers providing database, git, HTTP, and memory management capabilities
This collection exhibits pervasive quality gaps across all 23 tools. Descriptions are present but minimal (10-50 chars), falling well below the 194-char baseline for production tools. Input schemas are visible and typed, but parameter descriptions are sparse or generic. No output schemas are documented. Error handling is minimal (generic 'Error: {str(e)}' responses). Tools lack composition guidance, no chaining IDs returned, no pagination support despite potential for large result sets. Security concerns: database credentials loaded from config files, no input validation against SQL injection despite execute_query accepting arbitrary SQL. The server is STDIO-only, which is a hard transport limitation.
快速检查 URL 状态(只返回状态码)
清空所有记忆
删除记忆
执行 SQL 查询(只读 SELECT)
根据改动生成 commit message 建议
查看分支列表
查看当前分支
查看未提交的改动
Descriptions are uniformly minimal (10-50 chars), far below the 194-char production baseline. LLMs cannot determine when to select tools or understand prerequisites. Example: 'get_status' → '查看 Git 状态' (11 chars) lacks context on when to use it vs other git tools.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields. execute_query returns raw dict rows; get_tables returns newline-separated strings. Inconsistent formats waste tokens and invite parsing errors.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 43 | 2026-07-28+ | v2 |
查看最近的请求历史
查看提交历史
获取记忆
查看 Git 状态
获取表结构
获取所有表名
快速 GET 请求
发送 HTTP 请求(完整版)
列出所有配置的数据库
列出所有记忆
列出所有可访问的项目
快速 POST JSON 请求
保存记忆
测试数据库连接
快速测试本地 API(简化本地测试)
execute_query accepts arbitrary SQL strings with only a basic 'SELECT|WITH' prefix check. No parameterized queries, no prepared statements. Vulnerable to SQL injection if LLM is tricked into passing malicious payloads. Credentials loaded from plaintext config file.
No pagination support. execute_query caps results at 50 rows but returns no next_cursor or total_count. Large result sets will be truncated silently, and LLMs cannot iterate. get_log accepts a limit param but no offset/cursor.
Error responses are generic ('Error: {str(e)}') with no recovery guidance. LLMs cannot determine if an error is retryable, user-fixable, or fatal. No actionable error messages (e.g., 'Invalid status: got X, expected one of Y').