MCP (Model Context Protocol) server for OpenProject project management integration with project analysis, work package management, and report generation capabilities
This MCP server has moderate structural quality but significant gaps in descriptions, parameter documentation, and error handling. All 9 tools have explicit schemas and are properly registered in MCPToolManager.list_tools(), but descriptions are brief (mostly 10-30 characters in Chinese), parameters lack detail, and there is no visible error handling or output schema documentation. The server targets OpenProject integration with project/work package retrieval and report generation, but the shallow descriptions make it difficult for LLMs to determine when to use each tool and what each parameter controls. No tool annotations (readOnlyHint, destructiveHint, idempotentHint) are present. The code structure is clean (Pydantic, clean architecture), but tool quality lags behind production baselines.
评估项目风险
生成项目月报
使用模板生成报告
生成项目周报
获取指定项目的详细信息
获取所有项目列表
获取工作包列表
All tool descriptions are extremely brief (10-30 characters in Chinese) and lack context. They do not explain WHEN to use each tool, what prerequisites exist, or what the output contains. Per the rubric baseline of 194 chars average, these descriptions underperform significantly and hamper LLM tool selection.
Parameter descriptions are missing or minimal. For example, 'project_id' is described as '项目 ID' (Project ID) with no guidance on format, whether it is numeric or alphanumeric, or how to obtain it. 'template_data' in save_report_template has only '模板数据' (template data), no schema, no examples of valid structure, no guidance on required fields. Per the rubric, 100% of A+ tools have descriptions for every parameter.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | D | 53 | 2026-07-28+ | v2 |
| 2026-03-09 | F | 37 | 2024-11-05+ | v1 |
获取所有报告模板列表
保存报告模板
No output schemas are documented in the source code. Tools like generate_weekly_report and generate_monthly_report return reports, but the structure of the report (fields, types, pagination) is not specified. LLMs cannot reason about downstream tool chaining or data extraction without knowing the response schema. Per the rubric, 100% of A+ tools have documented return types.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) are present. The MCP spec (2026-07-28) supports tool annotations to mark which tools are read-only, destructive, or idempotent. This server does not use them, reducing clarity about side effects. save_report_template is destructive (modifies state) but has no annotation.
Date parameters in generate_weekly_report and generate_monthly_report lack validation guidance. Descriptions say '(YYYY-MM-DD)' and '(1-12)' in comments but not in the schema. The inputSchema has no format or pattern constraints. LLMs may pass invalid dates (e.g., 2024-13-01 or day 32). Per the rubric, numeric ranges and date formats should be enforced in schemas and documented.
No error handling or recovery guidance is visible in the tool definitions. There is no documentation of what happens if a project_id does not exist, if a date range is invalid, or if a report generation fails. Per the rubric pattern:recovery-guide, error responses should tell the LLM what to do next, but no such guidance is present.
The 'custom_data' parameter in generate_report_from_template is marked optional but has no type definition. It is described only as '自定义数据(可选)' (custom data, optional). What structure does it expect? What fields are allowed? Per the rubric, every parameter needs a description explaining what it controls and its expected format.
No pagination support is visible. Tools like get_projects and get_work_packages do not specify limit/offset or page parameters, nor do they document response size caps. Per the rubric baseline, tools returning lists should accept pagination parameters and return a total count. Unbounded lists risk context window exhaustion.