krt-jira-scribe
Pass
Audited by Gen Agent Trust Hub on May 12, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it retrieves and processes untrusted data from an external Jira instance (issue summaries, descriptions, and comments).
- Ingestion points: Searches and reads issues via
/rest/api/2/searchand/rest/api/2/issue/{key}(referenced inSKILL.mdandreferences/jira-api.md). - Boundary markers: None explicitly defined to separate Jira content from agent instructions.
- Capability inventory: Executes shell commands via
curlfor reading and writing data, and performs status transitions. - Sanitization: No explicit sanitization or validation of the retrieved Jira content is mentioned before it is displayed or used in prompts.
- [COMMAND_EXECUTION]: The skill provides explicit instructions to bypass privacy/security filters implemented by command wrappers like
rtkthat may mask sensitive environment variables. - Evidence:
SKILL.mdandreferences/jira-api.mdcontain instructions to use direct shell checks (e.g.,[[ -n "$JIRA_HOST" ]]) specifically because automated filters might hide these variables from the agent's view.
Audit Metadata