cron-runner

Warn

Audited by Gen Agent Trust Hub on Mar 22, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill employs shell-like operations such as renaming (mv) and deleting files within the local runtime directory to manage its processing queue.
  • [REMOTE_CODE_EXECUTION]: The orchestrator extracts and dispatches commands directly from the cron-actions-queue.jsonl file to the agent's router() system. This enables the execution of arbitrary agent tasks based on unvalidated data from an external file.
  • [PROMPT_INJECTION]: The skill provides an indirect prompt injection surface by ingesting and acting upon instructions from a shared queue file without sanitization.
  • Ingestion points: Data is read line-by-line from .claude/context/runtime/cron-actions-queue.jsonl (SKILL.md).
  • Boundary markers: No delimiters or protective instructions are used to separate task data from the agent's execution context.
  • Capability inventory: The skill can dispatch tasks, update system state, and perform file system manipulations (SKILL.md).
  • Sanitization: The processing logic only checks for JSON validity and does not inspect the contents of the tasks for safety or authorization.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 22, 2026, 04:50 PM