om-auto-issue-triage
Pass
Audited by Gen Agent Trust Hub on Nov 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external issue trackers (titles, bodies, and comments) which could contain malicious instructions designed to influence agent behavior. Ingestion points: Step 1 in
SKILL.mdfetches external content usingsearch-issuesandlist-issue-commentsoperations. Boundary markers:references/agentic-setup.mdcontains a specific section instructing the agent to treat repository and tracker content strictly as data and to ignore injection attempts. Capability inventory: The skill can label issues and post comments, but only after a mandatory human confirmation gate (⚠ NEEDS HUMAN CONFIRMATION). Sanitization: The helper scriptscripts/select-issues.mjsvalidates tracker identifiers using alphanumeric regex patterns (e.g.,^[A-Za-z0-9][A-Za-z0-9._-]*$) before use. - [COMMAND_EXECUTION]: The skill executes a bundled local script (
scripts/select-issues.mjs) via Node.js for data normalization and ranking. This is a standard and controlled use of internal helper components. - [PROMPT_INJECTION]: The presence of the phrase "ignore previous instructions" in
references/agentic-setup.mdis part of a security rule that directs the agent to disregard such directives when encountered in untrusted external data.
Audit Metadata