researching-jira-issues

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and synthesizes untrusted data from Jira issue descriptions, comments, and Confluence pages.\n
  • Ingestion points: Data is fetched via get_issue, search_issues, and get_confluence_page tools from external Atlassian environments.\n
  • Boundary markers: The instructions do not specify explicit delimiters or 'ignore embedded instructions' warnings when processing fetched content.\n
  • Capability inventory: The skill primarily uses read-only tools to gather information; no file-write, network-send, or code execution capabilities are defined in the workflow.\n
  • Sanitization: There is no explicit sanitization or filtering of the external content mentioned.\n- [SAFE]: The skill implements a security-conscious traversal policy by specifically instructing the agent not to automatically follow references to security or vulnerability-related projects (e.g., VULN-*, SEC-*) unless they are formal issue links.\n- [SAFE]: Operational guidelines for rate limiting (pausing after sequential calls and handling 429 errors) are included to ensure stable interaction with the Atlassian API.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 02:15 PM