security-reviewer
Security checks across malware telemetry and agentic risk
Overview
This is a transparent, instruction-only security-audit skill, but it includes powerful pentest, secrets, and cloud commands that should only be run on explicitly authorized targets.
Install/use this skill only if you intend to perform authorized security review work. Before letting the agent run Bash commands, define the exact codebase, systems, cloud accounts, and test window; require confirmation for active scans, exploitation checks, credential searches, or cloud/Kubernetes commands; and redact secrets from outputs.
VirusTotal
63/63 vendors flagged this skill as clean.
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 run against the wrong target, these commands could trigger alarms, overload services, or be interpreted as unauthorized testing.
These are active scanning and testing examples that can be disruptive or abusive outside an authorized penetration-test scope. They are purpose-aligned and the same artifacts include rules of engagement, so this is a usage note rather than a concern.
nmap -sV -p- target.com
sqlmap -u "http://target.com/search?q=test" --batch
for i in {1..1000}; do
curl https://api.target.com/login -d "user=test&pass=test"
doneUse this skill only with written authorization, defined targets, time windows, and explicit approval before intrusive Bash commands.
Running these examples in the wrong environment could expose private credentials or secrets in the agent conversation or report.
The reference material includes commands that can reveal credentials, shell-history secrets, or Kubernetes secrets. This is consistent with authorized security testing, but it touches high-impact account and secret boundaries.
# Credential harvesting grep -r "password" /home/*/ cat ~/.bash_history | grep -i "pass\|pwd\|secret" ... kubectl get secrets --all-namespaces
Limit secret and credential checks to approved systems and paths, avoid printing real secret values, and redact any sensitive output before sharing reports.
A CI pipeline copied from these examples could run changed third-party code in the future.
The CI/CD examples use third-party GitHub Actions, including a mutable branch reference to `@master` and broad version references. These are documentation examples rather than skill install behavior, but copying them directly can introduce supply-chain risk.
- uses: returntocorp/semgrep-action@v1 - uses: gitleaks/gitleaks-action@v2 - uses: aquasecurity/trivy-action@master
Pin CI actions and tools to reviewed versions or commit SHAs before adding them to production pipelines.