The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill uses the Bash tool to execute Python scripts with arguments like --brand "{brand}" and --competitors "{competitors}" which are directly interpolated from user-supplied parameters. This creates a high-risk command injection vulnerability where shell metacharacters (e.g., ;, |, or $(...)) in the user input could lead to arbitrary command execution on the host system.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection as it aggregates content from various external marketing and social media sources. \n
Ingestion points: External data is ingested from Google Ads, Meta, LinkedIn, email platforms, CRMs, and web pages via multiple scripts mentioned in Step 2. \n
Boundary markers: The instructions lack any requirement for delimiters or "ignore instructions" prompts when processing this external content. \n
Capability inventory: The skill uses Bash, Read, and Grep tools and performs file writes to brand data directories (~/.claude-marketing/) and user documents (~/Documents/). \n
Sanitization: No validation or sanitization process is described for the content pulled from these external sources before it is analyzed by the agent or included in reports.

campaign-audit