deferred-task-execution
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses a Bash script to manage a background timer. This is the primary intended functionality of the skill to allow the agent to defer tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided duration and time strings. The implementation is secure as it uses a helper script with strong input validation.
- Ingestion points: Natural language time and duration strings provided by the user in the agent chat (e.g., "in 5 minutes", "at 3pm").
- Boundary markers: Not explicitly present in instructions, but the script handles parsing internally.
- Capability inventory: The skill uses the
Bashtool to run thewait-until.shscript; the script utilizes standard system utilities likesleep,date, andprintf. - Sanitization: The
wait-until.shscript implements strict regex-based parsing (^([0-9]+)h(.*)$,^([0-9]{1,2}):([0-9]{2})$, etc.) to extract numeric values. This prevents malicious input from being passed to sensitive shell commands or arithmetic expansions, effectively mitigating command injection within the script's logic.
Audit Metadata