The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local Python scripts (scripts/dfs_client.py and scripts/report_publish.py) using shell commands. These commands interpolate user-provided values like <topic> and <report>.md, which presents a risk of command injection if the inputs are not strictly validated or escaped before execution.
[INDIRECT_PROMPT_INJECTION]: The skill fetches and processes content from external web pages (top 5 organic search results) using the webfetch tool. This exposes the agent to instructions hidden in third-party content.
Ingestion points: External web content retrieved via webfetch and keyword data from dfs_client.py (referenced in SKILL.md).
Boundary markers: No explicit delimiters or "ignore" instructions are provided to separate fetched content from the agent's internal logic.
Capability inventory: File system write access (saving reports to %SEO_REPORTS_DIR% or $SEO_REPORTS_DIR), environment variable access, and local script execution (python scripts/...).
Sanitization: No sanitization or filtering of the fetched web content is described before entity and relationship extraction.

semantic-seo