The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill's architecture relies on local markdown references and contained analysis scripts.- [COMMAND_EXECUTION]: The skill executes internal scripts scripts/lint_metadata.py and scripts/responsive_smoke_test.mjs. These tools perform targeted audits (metadata regex matching and site connectivity checks) and do not exhibit dangerous execution patterns like arbitrary code evaluation.- [DATA_EXFILTRATION]: The scripts/responsive_smoke_test.mjs script performs network GET requests using fetch() to verify the reachability of user-provided URLs. This is a core functional requirement for a web-auditing skill and does not involve the unauthorized transmission of sensitive data.- [SAFE]: Indirect prompt injection risks are minimal. While the skill ingests project files and external URL content for auditing, it only extracts metadata markers or connectivity status. Ingestion points (local source files and remote HTML) are handled without dynamic code execution or unsanitized prompt interpolation.

marketing-site-skill