The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions in SKILL.md require the execution of a local Python script: python campaign-tracker.py --brand {slug} --action list-campaigns. The contents of this script are not provided within the skill package, making its behavior unverified.
[DATA_EXPOSURE]: The skill accesses local filesystem paths to retrieve brand information, including ~/.claude-marketing/brands/{slug}/profile.json and various files within ~/.claude-marketing/brands/{slug}/guidelines/. While these are domain-specific paths, they contain sensitive brand and compliance data.
[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its data ingestion process.
Ingestion points: It reads brand profiles, restriction lists, and style guidelines (e.g., restrictions.md, channel-styles.md) from the local filesystem.
Boundary markers: The skill lacks explicit boundary markers or instructions to ignore potential commands embedded within the brand data files.
Capability inventory: The agent can execute subprocesses via the campaign-tracker.py script.
Sanitization: There is no evidence of sanitization or validation performed on the brand guidelines before they are used to influence the agent's output behavior.

aeo-geo