auto-todo

Pass

Audited by Gen Agent Trust Hub on Mar 14, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted requirement documents. Ingestion points: docs/requirements/ or user-provided paths (SKILL.md). Boundary markers: No delimiters defined for parsing Tier 2 or Tier 3 markdown content (SKILL.md). Capability inventory: The skill can write a todolist.md file (SKILL.md). Sanitization: No filtering of extracted content is performed before file generation.
  • [DATA_EXFILTRATION]: The skill reads project manifest files to detect tech stacks but includes explicit constraints forbidding the agent from reading sensitive files like .env, credentials, or files containing secrets (SKILL.md).
  • [SAFE]: The skill implements a mandatory Hard Gate requiring user approval of the task breakdown before file writing and includes safety features like atomic writes and file backups.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 14, 2026, 10:24 AM