Back to skill

Security audit

cron-sentinel

Security checks for vulnerabilities and agentic risk

Overview

Cron Sentinel is a transparent local helper for monitoring scheduled jobs; its command execution and state file are disclosed and aligned with that purpose.

Install only if you want local monitoring for recurring jobs. Review any generated crontab lines before adding them, and use --no-output-tail for jobs that may print credentials or sensitive data.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Rogue AgentSelf-Modification, Session Persistence
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (6)

YARA rule 'backdoor_persistence': Backdoor persistence with malicious payloads (shell commands, SSH key injection, hidden root users) [malware]

High
Category
YARA Match
Confidence
75% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

md
orrect across timezones and DST.

## The pattern to set up

The whole design is two scheduled entries:

1. **The wrapped job** - the real task, run through `wrap`, on its normal schedule.
2. **The watchdog** - `check` on a short schedule (e.g. every 30 min) that pipes its output to wherever the user gets notified.

`crontab` prints both lines. Show them to the user and let the user add them with `crontab -e`. Never run `crontab`, edit crontab files, or create scheduled tasks yourself. In OpenClaw, suggest the two scheduled tasks (the wrapped command, and a short-interval `check` whose output routes to their channel), show exactly what each would run, and create them only if the user explicitly confirms.

## How to help

1. **Setting up a new schedule:** ask for the command, how often it should run, and whether retries make sense (yes for anything network-dependent). Then produce the wrapped crontab line via `crontab`, and explain the watchdog line. Always set `--expect-every` - without

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · CHANGELOG.md (reported line 18)May include surrounding context.

md
### Changed
- Replaced the frontmatter `permissions` block (not part of ClawHub's schema, so it wasn't being read) with `metadata.openclaw.requires` and `envVars` declarations.
- SKILL.md and the description now state plainly that Sentinel never edits a crontab, creates scheduled tasks, uses the network, or requires elevated privileges, and that the agent must not create scheduled tasks without the user's explicit confirmation.
- Corrected the description, which previously said commands run "via the shell."
- `crontab` output now opens with a note that Sentinel doesn't install anything.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
88% confidence
Finding

The skill explicitly describes capabilities to execute local commands, read an environment variable, and read/write a state file, but it does not declare an explicit tool scope such as permissions or allowed-tools. That mismatch weakens policy enforcement and user transparency, making it easier for a powerful skill to run with broader-than-expected access.

Content

No source excerpt is available for this finding.

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · SKILL.md (reported line 73)May include surrounding context.

md
1. **The wrapped job** - the real task, run through `wrap`, on its normal schedule.
2. **The watchdog** - `check` on a short schedule (e.g. every 30 min) that pipes its output to wherever the user gets notified.

`crontab` prints both lines. Show them to the user and let the user add them with `crontab -e`. Never run `crontab`, edit crontab files, or create scheduled tasks yourself. In OpenClaw, suggest the two scheduled tasks (the wrapped command, and a short-interval `check` whose output routes to their channel), show exactly what each would run, and create them only if the user explicitly confirms.

## How to help

Session Persistence

Medium
Category
Rogue Agent
Confidence
85% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · cron_sentinel.py (reported line 357)May include surrounding context.

python
1. **The wrapped job** - the real task, run through `wrap`, on its normal schedule.
2. **The watchdog** - `check` on a short schedule (e.g. every 30 min) that pipes its output to wherever the user gets notified.

`crontab` prints both lines. Show them to the user and let the user add them with `crontab -e`. Never run `crontab`, edit crontab files, or create scheduled tasks yourself. In OpenClaw, suggest the two scheduled tasks (the wrapped command, and a short-interval `check` whose output routes to their channel), show exactly what each would run, and create them only if the user explicitly confirms.

## How to help

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · cron_sentinel.py (reported line 187)May include surrounding context.

python
try:
            # Argument list with shell=False: the command is executed directly,
            # never interpreted by a shell, so its text can't inject extra commands.
            proc = subprocess.run(
                args.command,
                shell=False,
                capture_output=True,