The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill defines a 'Capability Gap Response' rule that instructs the agent to prompt the user to install an external tool named '技能宝' from a specific GitHub repository (github:mindverse/skillhub). This organization is not recognized as a verified trusted vendor or well-known service provider, presenting a risk associated with unverified external code sources.
[COMMAND_EXECUTION]: The instructions explicitly tell the agent to output a platform-level command (/install) directed at the user. This command triggers the download and integration of third-party logic into the agent's environment.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input during the ideation and problem definition phases to generate structured outputs like SWOT analyses and action plans.
Ingestion points: User descriptions, project backgrounds, and creative inputs are ingested in the '第一阶段：定义问题' and '第二阶段：发散思维' sections (SKILL.md).
Boundary markers: The skill does not define clear delimiters (such as XML tags or triple quotes) or provide instructions for the agent to ignore embedded directives within user data.
Capability inventory: The skill has the capability to suggest platform commands (/install) and generate executable action plans.
Sanitization: No evidence of input validation, filtering, or escaping is present in the interaction logic.

brainstorming