The Agent Skills Directory

[SAFE]: The skill is a legitimate tool for cloud security management. All scripts and instructions align with its stated purpose of configuring and tuning WAF rules.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface (Category 8) because it processes untrusted data from web request logs.
Ingestion points: The scripts/agent.py file retrieves sampled request data via the get_sampled_requests method, and SKILL.md provides Athena SQL queries to analyze raw WAF logs.
Boundary markers: Absent. The logs are processed and displayed without specific delimiters to segregate untrusted data from instructions.
Capability inventory: The skill possesses the capability to modify cloud security infrastructure through AWS WAFv2 API calls and CLI commands.
Sanitization: Absent. The scripts output raw request details such as URIs and source IPs directly to the console for auditing purposes. This is expected behavior for a security analysis tool but noted as a potential injection vector if the output is consumed by other automated agents.
[EXTERNAL_DOWNLOADS]: The skill references documentation and libraries from trusted organizations including AWS, Microsoft, and Cloudflare. These resources are for legitimate configuration and integration purposes.

implementing-cloud-waf-rules