The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from multiple external sources, including Google search snippets, YouTube video comments, and ChatGPT web search responses. This content is presented to the agent for analysis, creating a surface for indirect prompt injection where malicious instructions embedded in the data could attempt to override agent behavior.
Ingestion points: SKILL.md documents tools like serp_organic_live_advanced (SERP results), serp_youtube_video_comments_live_advanced (YouTube comments), and ai_optimization_chat_gpt_scraper (ChatGPT responses).
Boundary markers: The instructions do not specify the use of delimiters or clear directives to ignore embedded instructions within the ingested data.
Capability inventory: The skill environment supports command execution (claude-seo run) and extensive network operations through the MCP server tools.
Sanitization: There is no evidence of content sanitization or validation for the external data before it is processed by the agent.
[COMMAND_EXECUTION]: The skill utilizes local shell scripts and a command runner for configuration and runtime management.
Evidence: SKILL.md includes instructions for running an installation script (./extensions/dataforseo/install.sh) and frequently calls a cost-tracking utility (claude-seo run dataforseo_costs.py). These patterns involve executing code within the local environment.

seo-dataforseo