devops-requirements

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external Azure DevOps work items, creating an indirect prompt injection surface.\n
  • Ingestion points: Data enters the agent context through dxs devops workitem, dxs devops workitems, dxs devops wiql, and dxs devops attachments commands in SKILL.md.\n
  • Boundary markers: There are no boundary markers or instructions to disregard embedded commands in the ingested work item fields (description, design).\n
  • Capability inventory: The skill possesses capabilities to execute local CLI commands (dxs devops, dxs report inspect) and write files to the artifact directory (<artifact_dir>/requirements/).\n
  • Sanitization: No sanitization, escaping, or validation of the fetched work item contents or attachment data is specified before interpretation.\n- [COMMAND_EXECUTION]: The skill executes various dxs CLI commands to fetch work items, run WIQL queries, list/download attachments, and inspect reports. These are standard vendor operations associated with the primary purpose of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 02:51 PM
Security Audit — agent-trust-hub — devops-requirements