The Agent Skills Directory

[SAFE]: No prompt injection or behavior override attempts were detected. The skill instructions reinforce grounding and simulate persona disclaimers.\n- [SAFE]: No evidence of unauthorized data exposure, exfiltration, or credential harvesting was identified.\n- [SAFE]: No obfuscation, hidden instructions, or dynamic code execution patterns were found in any of the skill files.\n- [SAFE]: The skill's data ingestion from project files and web searches was evaluated for Indirect Prompt Injection. Surface evaluation: Ingestion points include project context files and search results; Boundary markers are not explicitly defined; Capability inventory shows no shell or exfiltration tools; Sanitization is not specified. The surface is safe due to lack of exploitable capabilities.\n- [SAFE]: No persistence mechanisms, privilege escalation, or unsafe external downloads were identified.

marketing-council