The Agent Skills Directory

[SAFE]: The skill is composed of documentation, markdown templates, and Python data structures (using standard libraries like dataclasses and enum) intended for crisis management. No security threats were identified.
[INDIRECT_PROMPT_INJECTION]: The skill provides a framework for processing external, untrusted data such as social media content or news claims, which represents an attack surface for indirect prompt injection.
Ingestion points: Untrusted data enters the context via the Claim.text (SKILL.md) and CrisisMention.content (SKILL.md) structures.
Boundary markers: None are defined for the ingested text.
Capability inventory: The skill does not contain any dangerous capabilities; it lacks network access, file system operations, or subprocess/execution commands. Actions are limited to internal logging and status updates.
Sanitization: No explicit sanitization of external content is performed.
Assessment: The severity is SAFE because there are no exploitable capabilities present in the skill that could be triggered by malicious data.

crisis-communications