recipe-fullstack-build

Pass

Audited by Gen Agent Trust Hub on May 9, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill ingests untrusted content from the filesystem to drive its orchestration logic and sub-agent prompts.
  • Ingestion points: Reads and processes markdown task files from docs/plans/tasks/ (SKILL.md).
  • Boundary markers: Partially mitigated by appending a "Scope boundary for subagents" block to sub-agent prompts to limit the context of sub-operations (SKILL.md).
  • Capability inventory: Possesses the ability to execute git commit, delete files in the docs/plans/ directory, and invoke other specialized sub-agents (SKILL.md).
  • Sanitization: The skill does not explicitly describe sanitization or validation of the content within the task files before passing them to the sub-agents.
  • [COMMAND_EXECUTION]: Filesystem and Version Control Operations. The skill performs sensitive operations as part of its primary lifecycle.
  • Git Operations: The skill is instructed to execute git commit automatically upon receiving approval from quality-fixer sub-agents (SKILL.md).
  • File Deletion: The skill performs cleanup by deleting task files in docs/plans/tasks/ upon completion of the workflow (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
May 9, 2026, 10:40 PM