The Agent Skills Directory

[SAFE]: The skill is a legitimate tool for SEO analysis and contains no malicious code, obfuscation, or unauthorized data access patterns. It explicitly utilizes official Semrush authentication flows, avoiding the need for hardcoded API keys.
[EXTERNAL_DOWNLOADS]: The skill references official Semrush editorial articles and documentation for methodology. These are trusted vendor resources provided for guidance.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, specifically top-10 search results and user-provided URLs.
Ingestion points: User-provided keywords, database locations, and external ranking URLs.
Boundary markers: Instructions require that Semrush data, manual observations, and model judgment remain "visibly separate" in the output.
Capability inventory: The skill uses discovery and report tools (keyword_research, organic_research, domain_overview) but lacks shell-execution or file-system write permissions.
Sanitization: The skill employs a structured evidence object to ensure that external data is handled as data rather than instructions, and requires labeling of data sources to maintain context.

serp-analysis