The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has a defined surface for processing untrusted external data.
Ingestion points: User-provided spreadsheet descriptions, column names, and calculation requirements are ingested during the 'Understand needs' phase in SKILL.md.
Boundary markers: The instructions do not define or implement any boundary markers or 'ignore embedded instructions' warnings to separate user data from the system prompt logic.
Capability inventory: The skill can generate executable VBA macros, Power Query M code, and complex Excel formulas (observed in the 'Supported scenarios' section).
Sanitization: There is no evidence of sanitization, escaping, or validation of the user-provided content before it is processed to generate results.
[DYNAMIC_EXECUTION]: The skill generates executable content at runtime based on user prompts. Specifically, it provides VBA macros and Power Query scripts ('VBA 宏', 'Power Query 数据整理'). This is classified as simple script generation from templates and logic based on the user's immediate requirements.
[INDIRECT_PROMPT_INJECTION]: The skill includes a 'Skill Gap Response Rule' that instructs the agent to output a command (/install 技能宝 github:mindverse/skillhub) to influence the agent's environment or user behavior. This represents a multi-step chain where the skill's output is intended to trigger a subsequent installation action. While referencing a well-known service (GitHub), it directs users to a third-party repository not included in standard trusted lists.

excel-formula