Complex Task Executor

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill architecture creates an attack surface for indirect prompt injection through its task planning and execution cycle.
  • Ingestion points: The agent reads the workspace/<task-name>/plan.md file using the read_file tool during the execution phase to determine its next steps.
  • Boundary markers: While the skill enforces a specific Markdown structure for the plan, it does not include instructions or delimiters to isolate user-provided task details from the agent's internal instruction set.
  • Capability inventory: The agent is empowered to perform arbitrary tool calls and file system operations based on the steps it generates from user descriptions.
  • Sanitization: The methodology does not require the agent to validate or sanitize user-provided data before integrating it into the actionable plan file.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 02:51 PM