The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill executes several internal Python scripts, including approval-manager.py, posting-time-analyzer.py, hashtag-analyzer.py, and brand-voice-scorer.py. These are invoked from paths potentially determined by environment variables like ${CLAUDE_PLUGIN_ROOT}.
[COMMAND_EXECUTION]: The skill constructs shell commands by interpolating variables such as {slug}, {platform}, and {file} directly into script arguments. This pattern presents a risk of command injection if the variables, particularly those derived from user input or external files, are not strictly sanitized before execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data which could influence its behavior or the behavior of connected tools. 1. Ingestion points: User-provided 'Post content', 'Hashtags', 'Target platforms', and brand profile files located in the home directory (~/.claude-marketing/). 2. Boundary markers: No explicit delimiters or instructions are provided to the agent to treat interpolated content as untrusted data. 3. Capability inventory: The agent has the ability to execute local Python scripts via the shell and interact with multiple social media platforms through connected MCP servers. 4. Sanitization: There is no evidence of sanitization or escaping of the user-provided content before it is interpolated into shell command arguments or processed by analysis scripts.

schedule-social