T09 · Insecure Skill Coding Practices
- Location
SKILL.md:72- Finding
Unbounded Retries Can Cause Repeated or Duplicate Cron Job Execution
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This skill is coherent, but it asks agents to automatically rerun failed cron jobs without enough safeguards against repeated or duplicate side effects.
Install only for cron jobs that are known to be idempotent or have duplicate-suppression protections. Before adding the heartbeat integration, add explicit per-job opt-in, a cooldown using lastRunAtMs, retry limits, logging, and manual review for repeated failures.
SKILL.md:72Unbounded Retries Can Cause Repeated or Duplicate Cron Job Execution
The skill instructs agents to automatically rerun failed cron jobs based on error text without warning that reruns may repeat side effects such as duplicate messages, duplicate API writes, or repeated external actions. Because cron jobs can be non-idempotent, automatic recovery in this context increases the chance of unintended duplicate operations after transient failures.
The skill claims it prevents retry loops by checking lastRunAtMs, but the provided implementation never performs that check before rerunning failed jobs. In an automatic heartbeat-driven workflow, this discrepancy can cause repeated re-execution of the same failing cron job, especially during prolonged outages or recurring transient errors, leading to duplicate actions and unnecessary load.
No suspicious patterns detected.