backend-development-feature-development
Pass
Audited by Gen Agent Trust Hub on Jul 16, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it interpolates user-provided feature descriptions from the $ARGUMENTS variable directly into prompts for various subagents across all twelve phases of the development lifecycle in SKILL.md. While this is part of the intended orchestration functionality, the absence of boundary markers could allow malicious input to influence the behavior of downstream subagents.
- Ingestion points: User input provided via the $ARGUMENTS parameter is interpolated into prompts in SKILL.md (e.g., phases 1-12).
- Boundary markers: Absent. The prompts do not use delimiters such as XML tags, triple quotes, or specific markers to isolate the user-provided feature description from the agent's instructions.
- Capability inventory: The skill calls multiple specialized subagents via the Task tool (e.g., backend-architect, security-auditor, deployment-engineer, and data-engineer) which have significant implementation and deployment capabilities.
- Sanitization: Absent. User input is used directly in the prompts without evidence of validation, escaping, or filtering.
Audit Metadata