The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external 'artifacts' (plans, strategies, PRDs) which serves as an ingestion point for untrusted data.
Ingestion points: The artifact input field described in SKILL.md.
Boundary markers: The instructions do not specify the use of delimiters or explicit warnings to the agent to ignore instructions embedded within the artifact.
Capability inventory: The skill has no access to external tools, network operations, or file system modifications; its output is restricted to text generation.
Sanitization: No input sanitization or validation is implemented.
[SAFE]: The skill is entirely declarative and focuses on structured persona-based analysis. No obfuscation, data exfiltration patterns, or persistence mechanisms were detected. The skill's behavior matches its stated purpose of providing a critical review of user-provided content.

red-team-review