The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell script at ./extensions/banana/install.sh to install necessary components. The contents of this script are not provided, making its actions unverifiable.
[COMMAND_EXECUTION]: Multiple functional components of the skill rely on executing Python scripts via the Bash tool from the user's home directory (~/.claude/skills/seo-image-gen/scripts/), including presets.py, cost_tracker.py, and generate.py.
[EXTERNAL_DOWNLOADS]: The skill is sourced from a personal GitHub repository (AgriciDaniel/claude-seo) and requires the download of an external 'banana extension' for core functionality.
[REMOTE_CODE_EXECUTION]: The requirement to run an installation script from an external repository constitutes a potential remote code execution vector, as the script could perform unauthorized network operations or system modifications during the installation process.
[INDIRECT_PROMPT_INJECTION]: The skill is designed for SEO workflows which typically involve ingesting untrusted data from web pages. The agent has access to high-privilege tools such as Bash, Write, and WebFetch. There are no boundary markers or sanitization procedures defined to prevent instructions embedded in processed web content from influencing the agent's behavior.

seo-image-gen