The Agent Skills Directory

[SAFE]: The skill outlines a systematic approach for multi-agent threat modeling based on industry-standard reference architectures from the Cloud Security Alliance (CSA) and OWASP.
[SAFE]: No suspicious code patterns, credential access, persistence mechanisms, or obfuscation techniques were detected in the instructions.
[SAFE]: The requested tools (Read, Grep, Glob, Bash, WebFetch, Agent) are appropriate for an agent performing security analysis and documentation review tasks.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze external system descriptions and architectures. While this is an ingestion surface for untrusted data, the instructions do not facilitate instruction bypass or provide an exploitable path for malicious data to compromise the agent context.

multi-agentic-threat-model