The Agent Skills Directory

[SAFE]: The skill operates as a project configuration tool, interviewing the user to generate a structured brand brief. It uses local templates and does not require network access, external dependencies, or elevated privileges.\n- [PROMPT_INJECTION]: The skill identifies a surface for indirect prompt injection because it reads and processes data from an existing project file (.claude/brand-brief.md). This content is then used to influence the behavior of the agent and potentially other skills in the brand-protection workflow.\n
Ingestion points: The skill reads project-resident content from .claude/brand-brief.md during its initialization phase to determine existing brand settings.\n
Boundary markers: There are no explicit instructions or delimiters used to isolate project data from the agent's logic, which could allow maliciously crafted text in the brief to influence agent behavior.\n
Capability inventory: The skill has the capability to write to the project filesystem to update the brand brief.\n
Sanitization: No logic is present to sanitize or validate the content read from the project file before it is incorporated into the session context.

brand-voice