Skills
skills/deepeshbodh/human-in-loop/dag-operations/Gen Agent Trust Hub

dag-operations

Pass

Audited by Gen Agent Trust Hub on Feb 25, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses shell scripts to execute the hil-dag CLI utility for graph management operations. User-supplied arguments are safely handled via bash arrays to prevent shell injection.\n- [PROMPT_INJECTION]: The skill processes external configuration files (DAG and catalog JSONs) and descriptive metadata fields, creating an indirect prompt injection surface. This is a functional requirement for its workflow management role.\n
  • Ingestion points: DAG_PATH and CATALOG_PATH file inputs, and the --intent parameter in dag-assemble.sh.\n
  • Boundary markers: None present in the shell script wrappers.\n
  • Capability inventory: Limited to structural DAG operations via the hil-dag tool.\n
  • Sanitization: Employs standard bash array execution (\"${CMD[@]}\") to maintain command integrity.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 25, 2026, 02:12 AM