反无限循环守护
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This is a simple loop-guard skill with no credentials, network access, installation step, or persistence; the main caveat is that it may make an agent stop early based on fixed limits.
This skill appears safe to install from the provided artifacts. Use it as a lightweight guardrail, not a guaranteed loop-prevention system, and make sure its stop limits fit your task before relying on it.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If invoked with unsuitable thresholds, the agent may stop a task early even when more work could still be useful.
The skill explicitly introduces stopping conditions for the agent. This is the stated purpose, but it affects whether the agent continues or stops a task.
max_retries = 3 ... max_steps = 10 ... max_time_seconds = 300 ... break # 重复操作 → 停止
Use this skill when loop prevention is desired, and adjust or review thresholds for tasks that require many retries or long-running steps.
Users may overestimate how robust or well-tested the loop guard is.
The documentation makes precise performance and false-positive claims, while the provided tests are minimal and do not substantiate those claims.
检测延迟 | <10ms ... 内存开销 | <5MB ... 误报率 | <0.1%
Treat the metrics as unverified documentation claims unless the publisher provides stronger tests or benchmarks.