This MCP server exposes 7 enterprise risk analysis tools with significant quality gaps. All tools are READ_ONLY API wrappers with Chinese-language descriptions, but critical deficiencies in schema documentation, parameter typing, and output specification prevent confident LLM usage. While descriptions are present and moderately detailed, none of the tools declare explicit input schemas with typed parameters in the visible source code. Parameter lists are documented in docstrings but lack formal JSON Schema definitions. Output schemas are partially documented in docstrings but lack structured field typing. No error handling guidance, enum constraints, or recovery instructions are provided. The naming convention is consistent (risk_insight_* and enterprise_risk_*) but lacks action verbs (search_, get_, list_) that would clarify intent. Overall, this server reads as a thin HTTP wrapper with incomplete tool definition metadata.
该接口的功能是通过传入企业全称来查询某一家企业风险评分以及风险等级,并且列出详细的评分情况,用于企业入驻风险审核等场景。如果没有企业全称则先调取企业模糊查询工具获取企业全称。
该接口的功能是查询企业的动产抵押信息,提供有关动产抵押的详细数据,包括抵押的数量、列表信息以及相关主体和债权信息。可能的使用场景包括金融机构在进行信贷审批时评估企业资产负债情况,法律和金融分析师在制作企业信用报告时验证抵押信息,以及企业与合作伙伴在商讨业务合作时提供的资产担保状况核实。
该接口的功能是查询特定企业在法院公告中的相关信息,包括庭审地点、当事人、案由等细节,并以列表形式呈现。这一接口主要应用于法律、合规领域内的背景调查中,帮助律师、法务人员或企业合规团队获取目标企业涉及的历史及当前法律诉讼或公告信息,从而评估企业潜在法律风险。
该接口用于查询与给定企业相关的开庭公告信息,提供详细的庭审和公告细节。这一功能可以帮助法律从业者、企业管理人员或风险控制部门了解特定企业的法律诉讼情况及历史记录。场景包括企业尽职调查、市场竞争分析,以及企业内部风险管理,例如当涉及到企业收购或投资决策时,需要评估目标企业的法律状态与潜在风险。
企业模糊查询工具: 该接口的功能是根据提供的企业名称简称、人名、品牌、产品、岗位等关键词模糊查询相关企业列表。返回匹配的企业列表及其详细信息,用于模糊查询匹配企业全称,如果用户输入的不是企业全称则优先调用该接口。
No formal JSON Schema definitions for any tool. All parameter schemas and output schemas are documented in docstrings only, not in the @mcp.tool() decorator or tool registry. This means LLMs cannot introspect parameter types, constraints, or valid enums programmatically.
Enumerated parameters (keywordType: name|nameId|regNumber|socialCreditCode) are documented as text in docstrings but NOT defined as enum constraints in code. LLMs cannot automatically select valid options and must guess or hallucinate values.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-22 | F | 41 | 2026-07-28+ | v2 |
该接口的功能是查询企业的法律诉讼风险情况,输出包括法院公告、立案公告、开庭公告、裁判文书、被执行人、失信被执行人和限制高消费等信息,帮助用户评估企业的法律风险状况。该接口可能用于信贷风险评估、投资决策支持、企业合规审核、大数据分析中的舆情监测等场景,在这些场合中,获取企业的法律风险信息可以支持更全面的风险评估和决策过程。
该接口的功能是查询某一家企业在政府监管下的严重违法记录,包括严重违法的详细信息和处理状态。可能的使用场景包括:在金融贷款审批中,银行或金融机构使用该接口来判断借款企业的合规风险;在企业尽职调查中,投资者或合作伙伴通过此接口获取目标企业的合规历史;在招聘过程中,人力资源部门考察候选公司的背景和合法性;甚至在市场及监管机构内部用于对企业做出合规监督和处罚决策。
Tool names do not follow action-verb conventions (get_, search_, list_, create_, etc.). Names like 'risk_insight_fuzzy_search' and 'enterprise_risk_insight_score' are descriptive but ambiguous about intent. LLMs must infer from description alone.
Output schemas are documented in docstrings but not formalized as return type annotations or structured objects. For example, risk_insight_chattel_mortgage returns a dict with 10+ nested fields, but the response structure is not validated or typed in code.
No error handling or recovery guidance. If an API call returns an error (e.g., enterprise not found, API quota exceeded, network timeout), the tool returns a generic dict with 'error' key or a string. No guidance on whether the error is retryable, user-fixable, or fatal.
Pagination parameters (pageIndex, pageSize) are present in some tools but max limits are stated only in docstring text ('one page max 50 records'), not enforced as schema constraints. No total count or next_cursor returned to guide multi-page iteration.
API credentials (INTEGRATOR_ID, SECRET_ID, SECRET_KEY) are injected from environment variables, which is correct. However, the API endpoint URL is hardcoded in call_api(), and error responses may leak information about the backend API structure.