ai-automation-workflows

Pass

Audited by Gen Agent Trust Hub on Mar 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: Orchestrates workflows using the infsh CLI and complex Bash scripting including loops, conditionals, and background processes.
  • [PROMPT_INJECTION]: Vulnerable to indirect prompt injection in data_processing.sh and conditional_workflow.sh. External data from files ($(cat $file)) and command-line arguments ($1) are directly interpolated into LLM prompts.
  • Ingestion points: conditional_workflow.sh (CLI arguments), data_processing.sh (local file content in SKILL.md).
  • Boundary markers: Absent; no delimiters or instructions to ignore embedded commands are used.
  • Capability inventory: Shell access via Bash tool, network access via curl and infsh, and file system access (specified in SKILL.md).
  • Sanitization: Absent; input is used raw in the prompt string.
  • [EXTERNAL_DOWNLOADS]: Recommends installing additional skills from the inference-sh repository via npx skills add. These are vendor-provided resources from the author.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 18, 2026, 01:20 PM