The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions include scanning the local filesystem for blog-related files (markdown and HTML) using glob patterns. This is used to build a content inventory for the editorial calendar.
[DATA_EXFILTRATION]: It reads specific metadata fields from existing content to assess content decay. There is no evidence of this data being sent to unauthorized external endpoints.
[PROMPT_INJECTION]: The skill processes untrusted input from existing local files and external web search results, creating an indirect prompt injection surface. Evidence: 1. Ingestion points: existing blog files and WebSearch results; 2. Boundary markers: Absent; 3. Capability inventory: Subprocess calls or high-risk writes are absent in this specific workflow; 4. Sanitization: Absent. This risk is evaluated as safe given it is necessary for the skill's primary function.

blog-calendar