backend-development-feature-development
Pass
Audited by Gen Agent Trust Hub on Jun 21, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill interpolates the
$ARGUMENTSvariable directly into prompts for various sub-agents across all four phases. This pattern allows user input to blend with the sub-agent's instructions, potentially leading to prompt injection if the user provides malicious content. - Evidence:
Prompt: "Analyze feature requirements for: $ARGUMENTS. ..."in Phase 1, Step 1. - Evidence:
Prompt: "Design technical architecture for feature: $ARGUMENTS. ..."in Phase 1, Step 2. - [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface by passing untrusted data to sub-agents with significant capabilities without adequate protection.
- Ingestion points: User-provided feature descriptions via the
$ARGUMENTSvariable in SKILL.md. - Boundary markers: Absent. The variable is interpolated into string templates without delimiters (e.g., XML tags or clear separators) or instructions to the sub-agent to ignore embedded commands.
- Capability inventory: The skill orchestrates sub-agents via the
Tasktool with capabilities for backend implementation, security auditing, and infrastructure deployment. - Sanitization: No validation, escaping, or filtering of the user input is implemented before interpolation.
Audit Metadata