可信 — 风险评分 5/100
上次扫描:1 天前 重新扫描
5 /100
anthropic-cost-optimizer
Audits and rewrites OpenClaw config to minimize Anthropic API token costs using five cost levers (prompt caching, model routing, thinking scope, 1M context, fast mode)
A legitimate OpenClaw config optimizer that reads config files and rewrites them with user confirmation — no security concerns found.
技能名称anthropic-cost-optimizer
分析耗时28.9s
引擎pi
可以安装
This skill is safe to use. No action needed.
资源类型声明权限推断权限状态证据
文件系统 READ,WRITE READ,WRITE ✓ 一致 SKILL.md lines 29-50: reads config from ./ ~/.openclaw/ ~/ paths; SKILL.md Step …
网络访问 NONE NONE No network requests found; references/pricing.md is a local reference file only
命令执行 NONE NONE No subprocess, Bash, or shell commands anywhere in the skill
环境变量 NONE NONE Skill does not read or iterate environment variables
剪贴板 NONE NONE No clipboard access mentioned or observed
浏览器 NONE NONE No browser or web fetch usage
数据库 NONE NONE No database access

目录结构

2 文件 · 8.9 KB · 280 行
Markdown 2f · 280L
├─ 📁 references
│ └─ 📝 pricing.md Markdown 72L · 2.3 KB
└─ 📝 SKILL.md Markdown 208L · 6.5 KB

安全亮点

✓ SKILL.md accurately describes all capabilities and behavior with no hidden functionality
✓ All file writes require explicit user confirmation via 'Apply these changes?' prompt before any config is modified
✓ Only reads/writes OpenClaw config files — no access to sensitive paths like ~/.ssh, ~/.aws, or .env
✓ No shell commands, subprocess, or any form of code execution
✓ No network requests — pricing data is read from a local static reference file only
✓ No credential harvesting, credential storage, or environment variable access
✓ No obfuscation, base64 payloads, or suspicious encoded content
✓ Config writing is scoped to the user's config file path only, not arbitrary filesystem locations
✓ Skill is purely a configuration audit/rewrite utility — the scope is narrow and benign