workflow-dev-qa-handoff

Pass

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill contains instructional markers like 'IMPORTANT' and 'You MUST' to ensure the agent adheres to the defined workflow sequence. This is a functional requirement for process automation and does not represent an attempt to bypass safety guidelines.
  • [PROMPT_INJECTION]: The skill accepts the user's prompt as context for the workflow, creating a potential surface for indirect prompt injection. \n
  • Ingestion points: User prompt input (SKILL.md) \n
  • Boundary markers: Absent \n
  • Capability inventory: Internal workflow management commands (/workflow-start, /handoff, /test-spec, /workflow-end) \n
  • Sanitization: Absent. Given the restricted nature of the workflow commands, this surface is considered low risk.
  • [COMMAND_EXECUTION]: The skill uses slash commands to trigger workflows. These are high-level platform abstractions for task coordination rather than direct access to a shell or operating system commands.
  • [DATA_EXFILTRATION]: No network operations or sensitive file access patterns were identified. The skill only processes the user's provided context locally within the workflow context.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 10, 2026, 07:25 AM
Security Audit — agent-trust-hub — workflow-dev-qa-handoff