The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external document content (PDF and TeX files), which creates an inherent attack surface for indirect prompt injection. A maliciously crafted document in the corpus could attempt to misdirect the analysis subagents.
Ingestion points: Step 2 involves reading files from a user-provided <corpus-dir>.
Boundary markers: While the skill enforces a strict output schema for subagents, it does not explicitly define delimiters or boundary markers for the raw document content being read.
Capability inventory: The environment permits Read, Write, Bash, and Agent tool usage.
Sanitization: There is no explicit sanitization step for the content extracted from the corpus files before it is processed by the subagents.
[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a find command paired with wc -l to count files in the corpus directory. This is a functional requirement for session management and does not involve network operations or access to sensitive system directories.

voice-profile