subagent-driven-development

Pass

Audited by Gen Agent Trust Hub on Mar 25, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external implementation plans which presents an indirect prompt injection surface.
  • Ingestion points: Full text of tasks and plans are ingested in SKILL.md and passed to subagents in implementer-prompt.md and spec-reviewer-prompt.md.
  • Boundary markers: Markdown headers such as ## Task Description are used to structure the subagent prompts, but the system lacks explicit delimiters or instructions to ignore or sanitize instructions embedded within the untrusted plan content.
  • Capability inventory: The subagents are granted capabilities to modify the codebase, commit changes, and execute tests via the provided toolset.
  • Sanitization: No evidence of input validation, filtering, or sanitization of the implementation plan content was found.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 25, 2026, 12:59 PM