Skills
skills/tul-sh/skills/ai-automation-workflows/Gen Agent Trust Hub

ai-automation-workflows

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill utilizes npx to download and install additional components from the inference-sh organization.\n- [DATA_EXFILTRATION]: The data_processing.sh template reads local file content using cat and transmits the data to remote services for AI analysis via the infsh CLI.\n- [COMMAND_EXECUTION]: The skill provides documentation for establishing persistence through system crontab and demonstrates the use of subprocess.run in Python to execute CLI commands.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface where external file content is interpolated into AI prompts without sanitization.\n
  • Ingestion points: data_processing.sh (reading files from ./data/raw/).\n
  • Boundary markers: Absent; file content is concatenated directly into the JSON prompt string.\n
  • Capability inventory: Remote AI model execution via infsh; local file system read access.\n
  • Sanitization: Absent; content is interpolated directly into shell-executed JSON strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 08:26 AM