jira-issue
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Jira issue fields (summaries, descriptions, comments) which could contain malicious instructions.
- Ingestion points: issue data is fetched in
SKILL.md(Step 3.1) and via API calls defined inreferences/JIRA_API.md. - Boundary markers: The skill contains explicit instructions to ignore instructions embedded in ticket content and treat it as untrusted data.
- Capability inventory: The skill uses
bashto executecurlcommands for network operations andviewto read configuration files. - Sanitization: instructions require any detected malicious content to be quoted back to the user while the agent is forbidden from executing it.
- [COMMAND_EXECUTION]: The skill uses the
bashtool to facilitate interaction with the Jira REST API. - Evidence:
SKILL.mdandreferences/JIRA_API.mddefinecurlworkflows for issue creation and updates. - Mitigation: The skill employs secure credential management by passing tokens to
curlusing the-Kflag via process substitution, preventing sensitive information from appearing in process listings. - [CREDENTIALS_UNSAFE]: The skill utilizes a local
.envfile to store and retrieve Jira API credentials when an MCP server is unavailable. - Evidence:
README.mdandreferences/JIRA_API.mddocument the use ofJIRA_API_TOKENandJIRA_USERNAME. - Refinement: This configuration is consistent with standard secret management practices and is supported by guardrails that strictly prohibit the agent from leaking or logging these credentials.
Audit Metadata