Ai Efficiency Monitor

AI效率监控工具。监控AI任务执行效率，识别浪费模式（重复查询/过长上下文/无效重试/过度生成/低质量循环），生成优化建议和成本节省报告。支持OpenClaw日志解析、多模型成本计算、ASCII趋势图。

Security Scan

The skill's code broadly matches its stated purpose (parsing OpenClaw logs and producing efficiency reports), but there are packaging and instruction inconsistencies that merit caution before installing or running it (notably a missing top-level entrypoint and instructions that source a script from the user's home).

Purpose & Capability

The scripts perform log collection, pattern analysis, cost calculation and reporting which align with the description (OpenClaw log parsing, waste-mode detection, multi-model cost table). The cost and context tables in cost_calc.sh match the stated multi-model cost calculation feature. Minor inconsistencies: SKILL.md claims CLI entry aiemon at repo root but the package manifest does not include a top-level aiemon executable file (only src/*.sh). Version in SKILL.md (v1.1.0) differs from registry version (1.2.1). These point to packaging/maintenance sloppiness, not necessarily malice.

Instruction Scope

SKILL.md instructs the user to 'source ~/.openclaw/workspace/skills/utils/error-handler.sh' before use. Sourcing a script from the user's home runs arbitrary shell code in the current shell — this expands runtime authority beyond simple reading of logs. The collector explicitly reads ~/.openclaw/logs (LOG_DIR) and will parse any files there. While reading OpenClaw logs is coherent with purpose, the instruction to source a home-directory script (not declared in requires/config) and lack of any explicit consent/validation is a scope creep and a potential execution risk if that file is untrusted.

Install Mechanism

No install spec or remote downloads are declared; all code is present in the package (no network fetches). This is lower install risk. However, the package appears to be missing the advertised top-level aiemon entrypoint, so the provided scripts may not be runnable as-is without adding a wrapper or symlink — this is a packaging inconsistency rather than an install-time code download risk.

Credentials

The skill declares no required environment variables or credentials, which matches that it performs local log parsing and local calculations. It does, however, access user-local paths (~/.openclaw/logs and (per SKILL.md) ~/.openclaw/workspace/skills/utils/error-handler.sh) and will create/modify files under its data/ directory. Access to the user's OpenClaw logs is proportionate to an efficiency monitor, but those logs may contain sensitive data (API keys, requests, prompts). The instruction to source a user-home script increases sensitivity because it executes code from the user's environment.

Persistence & Privilege

always is false and the skill does not request permanent platform-level presence. It writes runtime output into a local data/ directory and does not modify other skills' configuration. No elevated privileges or always-on behavior requested.

What to consider before installing

This skill mostly does what it says (parses OpenClaw logs, finds waste patterns, computes costs), but there are several things to check before installing or running it: - Packaging: The SKILL.md and test scripts expect a top-level 'aiemon' CLI entrypoint, but that file is not present in the package manifest. Expect to create a small wrapper (e.g., a script that calls src/*.sh) or ask the provider for the missing entrypoint before use. - Review any sourced files: SKILL.md tells you to source ~/.openclaw/workspace/skills/utils/error-handler.sh. Sourcing executes code in your shell — inspect that file (and any file this skill would source) before running to ensure it is safe. - Logs contain sensitive data: The collector will read ~/.openclaw/logs. Ensure those logs do not contain secrets you don't want parsed/stored. Consider running in a sandbox or with a copy of logs first. - Required utilities: The scripts rely on standard Unix tools (grep, awk, bc, find, date, printf). Ensure these are available; tests may fail otherwise. - No network calls detected: The scripts do not contact external endpoints, which reduces immediate exfiltration risk — but the 'source' instruction could cause external behavior depending on the sourced file. Verify the sourced file and any wrapper scripts. - If you plan to use in production, run tests (tests/test_all.sh) in an isolated environment and inspect generated data/ files for unexpected content. If you cannot verify the missing aiemon entrypoint or the sourced error-handler, treat the package as untrusted until those are resolved. Given these packaging and instruction issues, proceed cautiously and verify the two main red flags (missing aiemon wrapper and the sourced home script) before trusting the skill with real logs or sensitive data.

AI Efficiency Monitor (aiemon)

版本: v1.1.0 | 创建者: 思捷娅科技 (SJYKJ) | 日期: 2026-03-16

🎯 功能概述

监控 AI 任务执行效率，识别浪费模式，生成优化建议和成本节省报告。

核心能力

• 数据收集: 解析 OpenClaw 日志，提取 Token 用量、API 调用、执行时间
• 浪费模式识别: 5 种内置模式检测（重复查询/过长上下文/无效重试/过度生成/低质量循环）
• 成本计算: 多模型价格表（智谱/DeepSeek/OpenAI 等）
• 报告生成: Markdown/JSON 格式，含 ASCII 趋势图
• 趋势分析: 日/周/月效率趋势

📦 安装

# 确保 error-handler 已就绪
source ~/.openclaw/workspace/skills/utils/error-handler.sh

🚀 CLI 使用

# 收集数据（从 OpenClaw 日志）
aiemon collect --source openclaw --days 7

# 分析浪费模式
aiemon analyze --patterns all

# 生成报告
aiemon report --format markdown --days 7
aiemon report --format json --days 30

# 计算成本
aiemon cost --model glm-5-turbo --tokens 10000

# 趋势分析
aiemon trends --days 30

# 列出内置模式
aiemon patterns list

📁 文件结构

ai-efficiency-monitor/
├── SKILL.md              # 本文档
├── aiemon                # CLI 入口
├── src/
│   ├── collector.sh      # 数据收集
│   ├── analyzer.sh       # 浪费模式分析
│   ├── reporter.sh       # 报告生成
│   ├── cost_calc.sh      # 成本计算（含模型价格表+上下文字典）
│   ├── patterns_list.sh  # 模式列表展示
│   └── patterns/
│       └── definitions.sh # 浪费模式定义（5种模式）
├── data/                 # 运行时数据
├── tests/test_all.sh
└── package.json

📄 许可证与版权声明

MIT License Copyright (c) 2026 思捷娅科技 (SJYKJ) 免费使用、修改和重新分发时，需注明出处。 出处：GitHub: https://github.com/zhaog100/xiaomili-skills | ClawHub: https://clawhub.com | 创建者: 思捷娅科技 (SJYKJ) 商业使用授权：个人免费 | 小微¥999/年 | 中型¥4,999/年 | 大型¥19,999/年