Cognitive Loop
Security checks across static analysis, malware telemetry, and agentic risk
Overview
This instruction-only skill is coherent and safety-conscious, but users should review the external package it recommends and understand that it stores reusable task memory.
Before installing, review the linked npm/GitHub package yourself, test in a sandbox, keep user approval for code changes or deployments, and monitor what gets saved under memory/.
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 used with file-editing, deployment, or command tools, the agent may attempt repeated fixes or tests as part of its workflow.
The skill describes autonomous retry, recovery, and repair behavior. This fits the stated cognitive-loop purpose, but could affect code or deployments if the user grants tools.
执行监控器 ... 自动重试、降级、异常恢复 ... 测试器 ... 自动修复
Keep approval gates for file changes, deployments, and other high-impact actions, especially during first use.
Installing the external package would run code outside this scan’s reviewed files.
The skill points users to an external npm package and GitHub repository that are not included in the reviewed artifact set. The instructions do tell users to inspect the source and package integrity.
npm install cognitive-agent ... git clone https://github.com/World-peace001/cognitive-agent.git
Inspect the repository/package, pin trusted versions, and test in a sandbox before installing or invoking the package.
Information from one task may be saved and influence later tasks.
The skill stores reusable execution patterns, lessons, successes, and knowledge. This is disclosed and path-scoped, but persistent memory can retain sensitive task details or be reused in future tasks.
经验自动保存到 `memory/` 目录 ... patterns/ ... lessons/ ... successes/ ... knowledge/
Avoid placing secrets or sensitive data in stored memory, and periodically review or clear the memory/ directory.