task-splitting

Pass

Audited by Gen Agent Trust Hub on Mar 22, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes the agent-team CLI for task creation as part of its core workflow. This functionality is restricted by a requirement for explicit user confirmation before any commands are executed, ensuring user control over the process.
  • [DATA_EXFILTRATION]: The skill reads user-specified local source documents to extract information for task decomposition. This access is scoped specifically to the provided files and is used to generate local task files in a sibling directory.
  • [PROMPT_INJECTION]: The skill processes untrusted input from source documents which presents a potential surface for indirect prompt injection. This risk is managed through a mandatory manual review step. Ingestion points: User-provided source documents in SKILL.md. Boundary markers: Instructions define the source document and user clarifications as the sole authority for task scope. Capability inventory: Access to the agent-team task create command. Sanitization: Requirement for explicit user confirmation before creating any task packages.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 22, 2026, 06:24 PM
Security Audit — agent-trust-hub — task-splitting