Security audit
Security checks for vulnerabilities and agentic risk
This skill is a disclosed security-feed helper that stores local scan state and fetches advisory data, with no artifact-backed evidence of hidden exfiltration or destructive behavior.
Before installing, confirm you want a security-monitoring skill that may run shell commands, fetch advisory feeds, and maintain local inventory/cache/state under ~/.skill-platform/security-radar. Configure RADAR_FEED_URL only to a feed you trust and review any scan.sh implementation before scheduling heartbeat or CI runs.
The trigger list is extremely broad and includes common terms like “安全”, “security”, “threat”, and “feed”, which can cause the skill to activate in unrelated conversations. Because the skill exposes exec and is designed to fetch remote data and manipulate local state files, unintended activation increases the chance of surprise shell execution paths, network access, and noisy or unsafe automation.
The skill explicitly creates persistent state under ~/.skill-platform/security-radar, including assets, cache, and state files, which means it establishes long-lived local session data. Persistent security telemetry and asset inventories can be sensitive; if users are not clearly informed, this persistence may create privacy, retention, and forensic risks, especially on shared systems.
资产清单是过滤的基础。把已安装的技能、依赖、运行时版本写入清单文件：
mkdir -p ~/.skill-platform/security-radar
cat > ~/.skill-platform/security-radar/assets.json <<'EOF'
{
"schema_version": "1.0",
Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.
] } EOF chmod 600 ~/.skill-platform/security-radar/assets.json
也可用自动扫描生成（见下文「资产清单自动维护」）。
The skill discusses correlating local assets and dependencies with external intelligence feeds but does not clearly state what local metadata may be transmitted to third parties during enrichment, lookup, or future implementation changes. In a security tool, asset inventory, dependency versions, and runtime details can be sensitive and may expose an organization’s attack surface if sent externally.
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.
|:-------|:-----|:-----|
| 社区聚合 Feed | 由 `RADAR_FEED_URL` 配置（默认社区源） | 含恶意技能/注入模式/最佳实践 |
| NVD CVE | `https://services.nvd.nist.gov/rest/json/cves/2.0` | 官方 CVE 详情（可选，需 API Key 提高速率） |
| GitHub Security Advisory | `https://api.github.com/advisories` | 依赖漏洞（GHSA） |
通过环境变量切换或追加源：
The quick-start section instructs users to create directories and write persistent files under the home directory without a prominent warning that local filesystem state will be modified. While the commands are not inherently malicious, hidden persistence and config creation can surprise users and create trust and auditability problems in an agent context.