The Agent Skills Directory

[COMMAND_EXECUTION]: The skill's workflow depends on the execution of multiple local Python scripts: posting-time-analyzer.py, hashtag-analyzer.py, brand-voice-scorer.py, approval-manager.py, and execution-tracker.py. These scripts are executed via subprocesses to process brand data and content, but their source code is not provided within the skill for verification.
[DATA_EXFILTRATION]: The skill reads sensitive information from the user's home directory, specifically within the ~/.claude-marketing/ path. This includes active brand configurations (_active-brand.json), detailed brand profiles (profile.json), and agency Standard Operating Procedures (SOPs). Accessing private configuration files in hidden directories is a data exposure risk.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by ingesting untrusted user content and local brand files, then processing them using scripts before final publication to external social media platforms.
Ingestion points: Processes user-provided post content, media asset URLs, and local brand profile data from ~/.claude-marketing/ (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions defined to separate untrusted content from the agent's internal logic.
Capability inventory: The skill possesses high-privilege capabilities including reading/writing local files in the home directory, executing Python scripts, and interacting with external social media APIs via MCP servers (SKILL.md).
Sanitization: The instructions do not specify any validation, escaping, or filtering of the user-provided content or local data before it is processed by the analyzer scripts or submitted for publication.

schedule-social