refine-story
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Jira issue summaries and descriptions via the getJiraIssue tool. Maliciously crafted content within a Jira ticket could attempt to redirect the agent's logic or influence the resulting refinement output.
- Ingestion points: getJiraIssue retrieves the issue key, summary, and full description from the Atlassian cloud in SKILL.md (Step 1).
- Boundary markers: The instructions do not define specific delimiters or 'ignore' warnings to isolate external issue content from the agent's instructions during the analysis phase.
- Capability inventory: The agent can search (searchJiraIssuesUsingJql) and modify (editJiraIssue) issue data in the Atlassian environment.
- Sanitization: No explicit sanitization, validation, or filtering of the fetched Jira content is performed before processing.
- Mitigation: The risk is mitigated by a mandatory human approval step (Step 5) where the user must explicitly confirm the refinement draft before any write operation is executed.
Audit Metadata