The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a Python script (intelligence-graph.py) located in the plugin root via subprocess. It passes parameters such as {slug} and {channel} which are derived from local configuration files and user requests. If these values are not properly sanitized, they could potentially be exploited for command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data from various JSON files and script outputs to synthesize reports and playbooks, which creates a surface for indirect prompt injection if the source data is compromised.
Ingestion points: Files located at ~/.claude-marketing/brands/, including profile.json and _manifest.json, and agency SOPs at ~/.claude-marketing/sops/.
Boundary markers: None identified in the skill instructions to distinguish between instructions and data.
Capability inventory: The skill has the ability to read local files and execute scripts through the Python interpreter.
Sanitization: The skill does not describe any specific validation or sanitization logic for the data ingested from the brand profiles or SOP files.

intelligence-report