fullstack-workspace-init

Warn

Audited by Gen Agent Trust Hub on Apr 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/init-workspace.py uses subprocess.run to execute shell commands, specifically calling python3 to run internal and external initialization scripts.
  • [REMOTE_CODE_EXECUTION]: The scripts/init-workspace.py script attempts to execute a file located at ~/.codex/skills/agent-folder-init/scripts/scaffold.py. This behavior involves running code from a path outside the skill's own directory structure, which introduces a dependency on the security and integrity of that external location.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by generating application code (NextJS components, NestJS services, etc.) based on a user-provided project 'brief'.
  • Ingestion points: The user-provided project description via the --brief argument in scripts/init-workspace.py and the intake workflow described in SKILL.md.
  • Boundary markers: There are no explicit boundary markers or 'ignore' instructions used to separate the user-provided brief from the generation logic.
  • Capability inventory: The skill possesses the capability to write files to the local file system and execute subprocesses (e.g., bun install).
  • Sanitization: The skill performs basic string transformations for naming but lacks robust sanitization of the logical content extracted from the user brief before it is interpolated into code templates.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 10, 2026, 09:45 AM