ticket-prepare
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external Jira tickets, including descriptions and attachments, which could contain malicious instructions intended to influence the agent's behavior.
- Ingestion points: Content is retrieved from Jira using available tools and from user discussions, then saved to local files like
ticket.mdand theassets/directory. - Boundary markers: The instructions define a specific structure for
ticket.mdand use numbering (D1, D2), but do not explicitly command the agent to ignore or escape instructions embedded within the imported ticket text. - Capability inventory: The skill uses file system tools to create directories and write files within the
.task/folder, and uses the Jira integration tool to fetch data. - Sanitization: There is no mention of sanitizing or escaping the imported content before it is written to the local workspace, where it may be processed by other skills in a pipeline.
Audit Metadata