Security audit
Security checks for vulnerabilities and agentic risk
This skill is a straightforward PRD analysis helper that reads local product and API reference material and writes a requirements document, with no hidden execution or persistence found.
Install this only in workspaces where the referenced competitor documents, API references, and optional Feishu destination are appropriate for the agent to use. The main thing to review is whether exposing those local paths and related skill names is acceptable for your environment.
The manifest description is entirely in Chinese and frames the skill's purpose and applicability only in that language. Elsewhere the document also specifies Chinese trigger phrases and output style, but nowhere offers the user a language/locale choice, which is a natural-language policy concern under the locale/language rule.
The skill explicitly enumerates internal skill names and repository paths, which discloses implementation details about available agent capabilities and local data layout. An adversarial user could use this information to steer the agent toward sensitive internal resources, craft follow-on prompts targeting those skills, or map the environment for broader prompt-level abuse.
开放平台Skill索引:
企迈 (qmai-api-fetcher): 32模块, 307接口
Skill路径: skills/qmai-api-fetcher/SKILL.md
数据路径: skills/qmai-api-fetcher/references/
美团 (meituan-api-fetcher): 6大类, 329接口
This line reveals another internal skill identifier and reference path, contributing to capability enumeration and environment mapping. In an agent ecosystem, such disclosure lowers the effort needed for an attacker to discover adjacent data sources and chain prompts toward unintended reads of internal documentation or API reference corpora.
数据路径: skills/qmai-api-fetcher/references/
美团 (meituan-api-fetcher): 6大类, 329接口
Skill路径: skills/meituan-api-fetcher/SKILL.md
数据路径: skills/meituan-api-fetcher/references/
客如云 (keruyun-api-fetcher): 20模块, 178接口
By naming the internal skill and data path for another platform integration, the document exposes the structure of the agent's local knowledge base. While not a direct compromise on its own, this materially assists reconnaissance and can increase the blast radius of prompt injection or data-exfiltration attempts against connected skills.
数据路径: skills/meituan-api-fetcher/references/
客如云 (keruyun-api-fetcher): 20模块, 178接口
Skill路径: skills/keruyun-api-fetcher/SKILL.md
数据路径: skills/keruyun-api-fetcher/references/
天财商龙 (tcsl-api-fetcher): 9模块, 99接口
This finding discloses the existence and location of an additional internal fetcher skill and its references directory, further expanding the attacker's map of the system. In context, the skill is designed to load local documents and API references, so revealing those exact targets makes the environment more dangerous by simplifying targeted prompt requests for internal content.
数据路径: skills/keruyun-api-fetcher/references/
天财商龙 (tcsl-api-fetcher): 9模块, 99接口 Skill路径: skills/tcsl-api-fetcher/SKILL.md 数据路径: skills/tcsl-api-fetcher/references/