Security audit
Security checks for vulnerabilities and agentic risk
This skill is transparent about monitoring and reactivating agents, but it can automatically steer other agents with broad scope and limited safeguards.
Install only if you intentionally want a tool that can resume or steer other agents. Prefer dry-run or target-specific use first, avoid cron-based auto-activation until you add allowlists and logging, and use thresholds that account for long-running quiet work.
The skill description is broad enough to trigger in many ordinary 'monitoring' or 'health check' contexts, which increases the chance of unintended activation. In an agentic environment, ambiguous activation criteria can cause this skill to run outside narrowly intended scenarios and steer agents without explicit operator intent.
This is a genuine autonomous decision-making pattern: the skill decides when an agent is 'stalled' and then issues a control message without human review. In context, that makes the behavior more dangerous because a simplistic time-based heuristic can misclassify long-running or quiet tasks and interfere with agent autonomy, task integrity, or system safety.
1. **Status Monitoring** - Real-time monitoring of agent runtime status
2. **Stall Detection** - Detecting "stalled" states where an agent has been unresponsive for over 5 minutes
3. **Automatic Activation** - Automatically sending activation messages to resume agent operation
## Core Capabilities
The skill explicitly describes automatically reactivating other agents but does not present a prominent warning that it performs autonomous control actions. That omission is dangerous because users may invoke or install the skill without realizing it can send commands to other agents, potentially disrupting tasks or overriding expected supervision boundaries.
This script is explicitly designed to make an autonomous operational decision: if an agent appears stalled, it can automatically send a steering/activation message without human approval. In an actual OpenClaw environment, this could interfere with agent behavior, revive intentionally paused agents, or create unintended control loops because the stall heuristic relies only on elapsed time and not task context or authorization checks.
Features:
1. Monitor agent runtime status
2. Detect "stalled" states where an agent exceeds the idle threshold
3. Automatically send activation messages to resume agent operation
Usage:
python monitor_agents.py --threshold 300 --auto-activate
No suspicious patterns detected.