The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes untrusted user data to guide agent behavior.
Ingestion points: External data enters the context through the --guidance command parameter and user responses during --interactive mode as described in SKILL.md.
Boundary markers: Absent. User-provided guidance is applied directly to adjust sub-agent emphasis without explicit delimiters or instructions to ignore embedded commands.
Capability inventory: The skill's capabilities are restricted to reading internal framework/rule files from $AIWG_ROOT and writing markdown reports to the .aiwg/marketing/competitive/ project subdirectory. It lacks network access, shell execution, or sensitive file access.
Sanitization: Absent. No input validation or filtering is performed on the provided guidance text.
[SAFE]: No evidence of malicious behavior was found. The skill does not attempt to access credentials, exfiltrate data, or execute remote code. All identified file operations are localized to the project workspace as intended by its marketing analysis purpose.

competitive-analysis