The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes local shell and Python scripts located at ~/.claude/skills/seo/scripts/preflight.sh and ~/.claude/skills/seo/scripts/keyword_research.py to manage authentication and verify API access. While these are localized to the skill's directory, they represent an execution of external code not contained within the main instruction file.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from an external source.
Ingestion points: Reads audit data from ~/.claude/skills/seo/output/<domain>-audit.json, which originates from the DataForSEO API.
Boundary markers: None identified; the agent is instructed to directly incorporate values like audit.executive_summary and issues table into the final report.
Capability inventory: The agent has Read and Write permissions, as well as the ability to execute shell scripts.
Sanitization: There are no instructions for sanitizing or escaping the content retrieved from the JSON file before formatting it into the markdown report.

seo-report