task-automation

Warn

Audited by Gen Agent Trust Hub on Mar 19, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions and code for establishing persistent execution mechanisms.\n
  • Evidence: Provides instructions for adding entries to crontab and mentions using systemd timers for recurring tasks.\n
  • Evidence: Includes a Python example script that utilizes the watchdog library to run as a persistent background observer for file system events.\n- [PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection through the processing of untrusted external data.\n
  • Ingestion points: The skill describes processes that read CSV files from user-accessible directories (~/data/incoming/) and consume JSON responses from a remote API (https://api.example.com/health).\n
  • Boundary markers: There are no instructions or delimiters provided to prevent the agent from interpreting instructions that might be embedded within the processed data.\n
  • Capability inventory: The automation examples include capabilities for file system modification (writing JSON, moving/deleting files) and making outbound network requests (urllib.request.urlopen).\n
  • Sanitization: While the skill performs basic validation of data structure (e.g., checking CSV headers), it does not implement sanitization of the data content itself.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 19, 2026, 01:40 AM