The Agent Skills Directory

[SAFE]: The skill follows security best practices for handling data bundles by explicitly documenting that its validation process does not verify factual accuracy, ownership, or safety of cited content.
[SAFE]: No evidence of prompt injection, data exfiltration, or obfuscated content was found in the skill instructions or configuration files.
[SAFE]: The skill uses localized MCP tools (e.g., seo_run_report) to perform its functions, avoiding the execution of arbitrary shell commands or the download of unverifiable external scripts.
[PROMPT_INJECTION]: The skill possesses a standard attack surface for indirect prompt injection as it ingests and processes external Markdown files provided in the 'files' parameter.
Ingestion points: External file content is passed to the seo_run_report tool via the files parameter in SKILL.md.
Boundary markers: Data is encapsulated within a structured JSON array of objects.
Capability inventory: The skill is restricted to calling the seo_list_reports, seo_describe_report, and seo_run_report tools for validation purposes.
Sanitization: While no explicit sanitization is performed by the skill itself, the instructions strictly limit the scope to structural validation and explicitly warn against interpreting structural success as a safety check.

okf-validate