The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted user data that is interpolated into the agent's context for strategic evaluation.
Ingestion points: User-provided marketing ideas, drafts, plans, and campaign descriptions passed via the --check argument or general conversation.
Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent to ignore instructions embedded within the processed content.
Capability inventory: The skill performs local file read and write operations (~/.claude-marketing/brands/{brand-slug}/goal-lock.json) using the agent's file system tools.
Sanitization: Absent. There is no instruction to escape, validate, or filter the external content before processing.
[COMMAND_EXECUTION]: The skill uses the agent's capabilities to read from and write to the local filesystem for state management.
Evidence: Instructions such as 'read the locked goal', 'write the file', and 'Append the verdict to history' involve operations on ~/.claude-marketing/brands/{brand-slug}/goal-lock.json.
The use of {brand-slug} in the file path without explicit validation or sanitization instructions presents a potential path traversal risk if the user provides a malicious brand name.

goal-filter