The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes a local Python script (dm-status.py) located within the plugin's root directory. It uses environment variables like ${CLAUDE_PLUGIN_ROOT} to resolve paths and passes user-supplied arguments such as brand slugs or section names to the script. This is standard behavior for an orchestration-layer skill.
[INDIRECT_PROMPT_INJECTION]: The skill reads and processes brand profiles, engagement logs, and marketing insights from the local filesystem (~/.claude-marketing/), which are then displayed to the agent. This data could potentially contain instructions intended to influence the agent's behavior.
Ingestion points: Brand JSON files and engagement data located in ~/.claude-marketing/brands/ and $CLAUDE_PLUGIN_DATA paths.
Boundary markers: Absent. The data is presented in a formatted snapshot without explicit delimiters to warn the agent about embedded instructions.
Capability inventory: The skill has the ability to execute shell commands via Bash and read files using Read, Grep, and Glob tools.
Sanitization: The output from the status script is passed directly to the agent context without sanitization or filtering.

status