The Agent Skills Directory

PROMPT_INJECTION (HIGH): The skill design exhibits a high-risk attack surface for indirect prompt injection by combining data ingestion with exploitable capabilities.
Ingestion points: External content enters the agent context through IncidentTimeline.add_event (description/source fields), LessonsLearned.set_summary, and Incident.add_action as documented in SKILL.md.
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the usage examples or documentation.
Capability inventory: The skill includes IncidentTimeline.export_csv(filepath), which allows the agent to write data to the file system, and report generation methods that produce structured text for downstream use.
Sanitization: There is no evidence of input validation, escaping, or filtering of external content before it is processed or stored.
NO_CODE (LOW): The referenced implementation script ir_utils.py is not included in the skill package, which limits the analysis to the provided interface documentation.

incident-response