OpenClaw Cron Guardrails

• Outcome: Review
• Latest audit: May 29, 2026 at 9:24 PM UTC
• Version:
  0.1.5

Overview

This cron helper is mostly transparent, but it normalizes recurring session prompt injection and can create or edit scheduled agent jobs, so users should review it carefully before installing.

Install only if you intentionally want a helper that can manage OpenClaw cron jobs and recurring agent actions. Review any job before applying it, avoid using the session/thread prompt-injection patterns, and only use webhook delivery with trusted endpoints and non-sensitive output.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Behavioral AST: exec() Call, eval() Call, Dynamic Import
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (8)

subprocess module call

Medium

• Category: Dangerous Code Execution

Content

print(json.dumps({"ok": True, "mode": "dry-run", "validation": validation, "render": rendered}, ensure_ascii=False, indent=2))
        raise SystemExit(0)

    proc = subprocess.run(command, shell=True, text=True, capture_output=True)
    result = {
        "ok": proc.returncode == 0,
        "mode": "apply",

• Confidence: 99% confidence
• Finding: proc = subprocess.run(command, shell=True, text=True, capture_output=True)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The activation guidance tells the agent to use this skill whenever a user wants any scheduled or repeated action, even if they never mention cron, and the metadata also covers broad natural-language reminders, recurring nudges, and repeated prompt injection. That broad trigger surface can cause the skill to engage in ordinary conversation and steer the agent toward file/script/shell-backed cron workflows unnecessarily, which raises the risk of unintended job creation, misrouting, or repeated prompt injection into a session/thread.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The activation guidance is intentionally broad enough to trigger on common natural-language scheduling phrases, which can cause the skill to activate outside a tightly bounded cron-management context. In an agentic system, overly broad routing increases the chance of misclassification and unintended execution of scheduled actions, including recurring nudges or thread/session injections, making this a real security-relevant issue even if not overtly malicious.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The webhook pattern explicitly routes cron output to an external HTTP endpoint, but the guidance does not warn that scheduled jobs may transmit prompts, results, or derived sensitive data outside the local/chat environment. In a scheduling skill, this is materially risky because users may copy the pattern for internal jobs without realizing they are creating an automated exfiltration path to a third-party service.

Missing User Warnings

Medium

• Confidence: 86% confidence
• Finding: The script performs immediate shell execution when --apply is supplied, without an additional confirmation or warning at the dangerous step. In a scheduling/agent skill context, this increases the chance that a user or upstream tool unintentionally causes execution of a sensitive cron-creation command or a maliciously altered rendered command.

Ssd 2

Medium

• Confidence: 98% confidence
• Finding: The example explicitly models 'prompt injection' against the current session agent as a valid schedulable intent, which normalizes an adversarial action into supported workflow behavior. In a cron/scheduled-actions skill, this is especially dangerous because it enables repeated, automated injection attempts against an active agent/session, increasing persistence and blast radius beyond a one-off unsafe prompt.

Ssd 2

Medium

• Confidence: 99% confidence
• Finding: The usage example repeats an attack-phrased command instructing prompt injection into the current session agent, effectively providing copy-pastable guidance for misuse. Because this skill is specifically for creating and routing cron jobs, documenting this as normal usage lowers operator caution and can directly facilitate recurring adversarial actions in live sessions.

Ssd 2

Medium

• Confidence: 91% confidence
• Finding: The parser intentionally recognizes phrases like 'nudge this session', 'push this thread', and 'inject prompt' as session-injection intent, which can normalize or enable repeated prompt delivery into the current conversation context. In a cron/scheduled-action skill, that is security-relevant because misclassification or permissive handling of these requests can lead to unauthorized persistence, repeated prompt injection, or routing of automated content into an active session/thread.

VirusTotal

66/66 vendors flagged this skill as clean.