{ "schemaVersion": "1.0", "item": { "slug": "oc-cost-analyzer", "name": "OC Cost Analyzer", "source": "tencent", "type": "skill", "category": "数据分析", "sourceUrl": "https://clawhub.ai/dagangtj/oc-cost-analyzer", "canonicalUrl": "https://clawhub.ai/dagangtj/oc-cost-analyzer", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/oc-cost-analyzer", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=oc-cost-analyzer", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "CHANGELOG.md", "README.md", "SECURITY.md", "SKILL.md", "package.json", "scripts/cost_analyzer.js" ], "primaryDoc": "SKILL.md", "quickSetup": [ "Download the package from Yavira.", "Extract the archive and review SKILL.md first.", "Import or place the package into your OpenClaw setup." ], "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run." } ] }, "sourceHealth": { "source": "tencent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-23T16:43:11.935Z", "expiresAt": "2026-04-30T16:43:11.935Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard", "contentDisposition": "attachment; filename=\"4claw-imageboard-1.0.1.zip\"", "redirectLocation": null, "bodySnippet": null }, "scope": "source", "summary": "Source download looks usable.", "detail": "Yavira can redirect you to the upstream package for this source.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/oc-cost-analyzer" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] }, "downloadPageUrl": "https://openagent3.xyz/downloads/oc-cost-analyzer", "agentPageUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent", "manifestUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent.json", "briefUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent.md" }, "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Then review README.md for any prerequisites, environment setup, or post-install checks. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "OpenClaw Cost Optimizer", "body": "专为 OpenClaw 用户设计的成本分析和优化工具。通过分析 session logs，识别高消耗场景，给出可执行的优化建议。" }, { "title": "1. 成本分析", "body": "读取 session logs，统计 token 使用和成本\n按模型、会话、时间维度分析\n识别高成本会话和异常消耗" }, { "title": "2. 场景识别", "body": "长对话检测: 超过 50k tokens 的会话\n频繁 Cron: 每天超过 10 次的定时任务\n大 Context: 平均输入超过 30k tokens\n昂贵模型: 使用高成本模型处理简单任务" }, { "title": "3. 优化建议", "body": "模型降级策略（Opus → Sonnet → DeepSeek）\nContext 压缩方案\nCron 频率调整\n本地模型使用建议\n预计节省金额" }, { "title": "生成完整分析报告（推荐）", "body": "node scripts/cost_analyzer.js analyze\n\n默认分析最近 7 天，生成详细报告保存到 ~/.openclaw/workspace/memory/cost-analysis-report.md\n\n指定天数:\n\nnode scripts/cost_analyzer.js analyze 30 # 分析最近 30 天" }, { "title": "快速查看今日成本", "body": "node scripts/cost_analyzer.js quick\n\n输出示例:\n\n📊 今日成本快览:\n 总成本: $2.45\n 会话数: 12\n 平均: $0.204/会话" }, { "title": "报告示例", "body": "# OpenClaw 成本分析报告\n\n生成时间: 2026-02-26 15:30:00\n\n## 📊 总览\n\n- 总会话数: 45\n- 总输入 tokens: 1,234,567\n- 总输出 tokens: 456,789\n- 总成本: $15.67\n- 平均每会话: $0.348\n\n## 🤖 模型使用统计\n\n### claude-opus-4-6\n- 会话数: 30\n- 输入: 890,123 tokens\n- 输出: 345,678 tokens\n- 成本: $12.34\n\n### claude-sonnet-4-20250514\n- 会话数: 15\n- 输入: 344,444 tokens\n- 输出: 111,111 tokens\n- 成本: $3.33\n\n## 💰 高成本会话 (Top 5)\n\n- Session: a1b2c3d4...\n - 模型: claude-opus-4-6\n - Tokens: 125,000\n - 成本: $3.45\n - 消息数: 25\n\n## 💡 优化建议\n\n### 1. 🔴 模型降级：yunyi-claude/claude-opus-4-6\n\n**问题**: 该模型成本较高 ($45/M tokens)，已使用 30 次会话\n\n**建议**: 对于简单任务使用 Sonnet 或 DeepSeek，复杂任务才用 Opus\n\n**预计节省**: $9.87\n\n**操作**: `openclaw models set yunyi-claude/claude-sonnet-4-20250514`\n\n### 2. 🟡 长对话检测\n\n**问题**: 发现 3 个长对话，最长 125,000 tokens\n\n**建议**: 超过 50k tokens 时开启新会话，避免 context 累积\n\n**预计节省**: $1.04\n\n**操作**: 手动开启新会话或设置 context 限制\n\n### 3. 🔴 Context 过大\n\n**问题**: 平均每次会话输入 27,435 tokens，可能加载了过多文件\n\n**建议**: 优化 AGENTS.md、SOUL.md，移除不必要的内容；使用 lazy loading\n\n**预计节省**: $6.27\n\n**操作**: 参考 openclaw-token-optimizer skill 的 context_optimizer\n\n**总预计节省**: $17.18" }, { "title": "策略 1: 模型分级使用", "body": "根据任务复杂度选择模型:\n\n任务类型推荐模型成本简单查询、文件读取local/qwen2.5:7b免费日常对话、代码编写claude-sonnet-4$3/M复杂推理、架构设计claude-opus-4-6$45/M备用/降级deepseek-chat$0.02/M\n\n操作:\n\n# 临时切换\nopenclaw models set yunyi-claude/claude-sonnet-4-20250514\n\n# 设置 fallback\nopenclaw models fallbacks add deepseek/deepseek-chat" }, { "title": "策略 2: Context 优化", "body": "问题: 默认加载所有 context 文件（SOUL.md, AGENTS.md, TOOLS.md, MEMORY.md, docs/**/*.md），可能 50k+ tokens\n\n解决方案:\n\n精简 AGENTS.md\n\n移除冗余说明\n合并重复规则\n使用简洁语言\n\n\n\nLazy Loading\n\n简单任务只加载 SOUL.md + IDENTITY.md\n复杂任务按需加载相关文档\n参考 openclaw-token-optimizer 的 context_optimizer\n\n\n\n定期清理\n# 清理旧 memory logs\nfind ~/.openclaw/workspace/memory -name \"2026-*.md\" -mtime +30 -delete" }, { "title": "策略 3: Cron 优化", "body": "识别高频 Cron:\n\nopenclaw cron list\n\n优化方案:\n\n非关键任务降低频率（每小时 → 每 4 小时）\n使用更便宜的模型执行 cron\n合并多个小任务为一个批处理\n\n示例:\n\n# 修改 cron 频率\nopenclaw cron edit \n\n# 为 cron 指定模型\nopenclaw cron add --model local/qwen2.5:7b \"0 */4 * * *\" \"检查系统状态\"" }, { "title": "策略 4: 会话管理", "body": "长对话问题: Context 累积导致每次请求都携带完整历史\n\n解决方案:\n\n超过 50k tokens 时主动开启新会话\n使用 /new 命令清空 context\n重要信息保存到 MEMORY.md，不依赖会话历史" }, { "title": "策略 5: 启用本地模型", "body": "适用场景:\n\n文件读取、简单查询\n开发测试\n离线工作\n\n设置:\n\n# 确保 Ollama 运行\nollama serve\n\n# 拉取模型\nollama pull qwen2.5:7b\n\n# 切换到本地模型\nopenclaw models set local/qwen2.5:7b" }, { "title": "成本基准", "body": "基于实际使用数据的成本参考:\n\n使用模式每日会话数平均 tokens/会话每日成本每月成本轻度使用5-1020k$0.50-1.00$15-30中度使用20-3030k$2.00-4.00$60-120重度使用50+40k$8.00-15.00$240-450优化后50+15k$3.00-6.00$90-180\n\n优化目标: 重度使用场景下节省 50-60% 成本" }, { "title": "每日成本检查（推荐）", "body": "添加 cron 任务，每天生成报告:\n\nopenclaw cron add \"0 9 * * *\" \"node ~/.openclaw/workspace/skills/openclaw-cost-optimizer/scripts/cost_analyzer.js quick\"" }, { "title": "每周深度分析", "body": "openclaw cron add \"0 10 * * 1\" \"node ~/.openclaw/workspace/skills/openclaw-cost-optimizer/scripts/cost_analyzer.js analyze 7\"" }, { "title": "成本告警", "body": "在 AGENTS.md 中添加规则:\n\n## 成本监控\n- 每日成本超过 $5 → 立即告警\n- 单次会话超过 $1 → 记录并分析\n- 每周生成成本报告" }, { "title": "openclaw-token-optimizer", "body": "本 skill 专注成本分析和报告\ntoken-optimizer 提供底层优化工具（context_optimizer, model_router）\n配合使用效果最佳" }, { "title": "openclaw-doctor", "body": "doctor 检查系统健康\ncost-optimizer 检查成本健康\n一起使用确保系统稳定且经济" }, { "title": "数据来源", "body": "Session logs: ~/.openclaw/agents/main/agent/sessions/*.jsonl\n分析最近 N 天的日志（默认 7 天）\n提取 inputTokens, outputTokens, model, sessionId" }, { "title": "成本计算", "body": "基于各模型官方定价:\n\nmodelCosts: {\n 'yunyi-claude/claude-opus-4-6': { input: 15, output: 75 },\n 'yunyi-claude/claude-sonnet-4-20250514': { input: 3, output: 15 },\n 'deepseek/deepseek-chat': { input: 0.014, output: 0.028 },\n 'local/qwen2.5:7b': { input: 0, output: 0 }\n}" }, { "title": "阈值配置", "body": "thresholds: {\n longConversation: 50000, // tokens\n highContextSession: 30000, // tokens\n frequentCron: 10, // 每天次数\n expensiveModel: 10 // USD per 1M tokens\n}\n\n可在脚本中修改以适应个人需求。" }, { "title": "安全性", "body": "✅ 纯本地运行\n\n无网络请求\n无外部依赖\n不执行子进程\n数据不离开本机\n\n✅ 只读分析\n\n只读取 session logs\n不修改任何配置\n不执行任何操作\n建议需手动执行" }, { "title": "问题: 未找到 session logs", "body": "原因: logs 目录不存在或路径错误\n\n解决:\n\nls ~/.openclaw/agents/main/agent/sessions/\n\n如果目录不存在，说明还没有会话记录。" }, { "title": "问题: 成本计算不准确", "body": "原因: 模型定价可能更新\n\n解决: 编辑 scripts/cost_analyzer.js，更新 modelCosts 对象。" }, { "title": "问题: 报告为空", "body": "原因: 指定天数内没有日志\n\n解决: 增加分析天数\n\nnode scripts/cost_analyzer.js analyze 30" }, { "title": "v1.0.0 (2026-02-26)", "body": "初始版本\n支持 session logs 分析\n识别 5 大高消耗场景\n生成优化建议和成本报告\n纯 Node.js 实现，无外部依赖" }, { "title": "贡献", "body": "欢迎提交 Issue 和 PR:\n\n新的优化策略\n更准确的成本计算\n更多模型支持\n报告格式改进" }, { "title": "许可", "body": "MIT License\n\n让 OpenClaw 更经济，让 AI 更可持续！ 💰✨" } ], "body": "OpenClaw Cost Optimizer\n\n专为 OpenClaw 用户设计的成本分析和优化工具。通过分析 session logs，识别高消耗场景，给出可执行的优化建议。\n\n核心功能\n1. 成本分析\n读取 session logs，统计 token 使用和成本\n按模型、会话、时间维度分析\n识别高成本会话和异常消耗\n2. 场景识别\n长对话检测: 超过 50k tokens 的会话\n频繁 Cron: 每天超过 10 次的定时任务\n大 Context: 平均输入超过 30k tokens\n昂贵模型: 使用高成本模型处理简单任务\n3. 优化建议\n模型降级策略（Opus → Sonnet → DeepSeek）\nContext 压缩方案\nCron 频率调整\n本地模型使用建议\n预计节省金额\n快速开始\n生成完整分析报告（推荐）\nnode scripts/cost_analyzer.js analyze\n\n\n默认分析最近 7 天，生成详细报告保存到 ~/.openclaw/workspace/memory/cost-analysis-report.md\n\n指定天数:\n\nnode scripts/cost_analyzer.js analyze 30 # 分析最近 30 天\n\n快速查看今日成本\nnode scripts/cost_analyzer.js quick\n\n\n输出示例:\n\n📊 今日成本快览:\n 总成本: $2.45\n 会话数: 12\n 平均: $0.204/会话\n\n报告示例\n# OpenClaw 成本分析报告\n\n生成时间: 2026-02-26 15:30:00\n\n## 📊 总览\n\n- 总会话数: 45\n- 总输入 tokens: 1,234,567\n- 总输出 tokens: 456,789\n- 总成本: $15.67\n- 平均每会话: $0.348\n\n## 🤖 模型使用统计\n\n### claude-opus-4-6\n- 会话数: 30\n- 输入: 890,123 tokens\n- 输出: 345,678 tokens\n- 成本: $12.34\n\n### claude-sonnet-4-20250514\n- 会话数: 15\n- 输入: 344,444 tokens\n- 输出: 111,111 tokens\n- 成本: $3.33\n\n## 💰 高成本会话 (Top 5)\n\n- Session: a1b2c3d4...\n - 模型: claude-opus-4-6\n - Tokens: 125,000\n - 成本: $3.45\n - 消息数: 25\n\n## 💡 优化建议\n\n### 1. 🔴 模型降级：yunyi-claude/claude-opus-4-6\n\n**问题**: 该模型成本较高 ($45/M tokens)，已使用 30 次会话\n\n**建议**: 对于简单任务使用 Sonnet 或 DeepSeek，复杂任务才用 Opus\n\n**预计节省**: $9.87\n\n**操作**: `openclaw models set yunyi-claude/claude-sonnet-4-20250514`\n\n### 2. 🟡 长对话检测\n\n**问题**: 发现 3 个长对话，最长 125,000 tokens\n\n**建议**: 超过 50k tokens 时开启新会话，避免 context 累积\n\n**预计节省**: $1.04\n\n**操作**: 手动开启新会话或设置 context 限制\n\n### 3. 🔴 Context 过大\n\n**问题**: 平均每次会话输入 27,435 tokens，可能加载了过多文件\n\n**建议**: 优化 AGENTS.md、SOUL.md，移除不必要的内容；使用 lazy loading\n\n**预计节省**: $6.27\n\n**操作**: 参考 openclaw-token-optimizer skill 的 context_optimizer\n\n**总预计节省**: $17.18\n\n优化策略详解\n策略 1: 模型分级使用\n\n根据任务复杂度选择模型:\n\n任务类型\t推荐模型\t成本\n简单查询、文件读取\tlocal/qwen2.5:7b\t免费\n日常对话、代码编写\tclaude-sonnet-4\t$3/M\n复杂推理、架构设计\tclaude-opus-4-6\t$45/M\n备用/降级\tdeepseek-chat\t$0.02/M\n\n操作:\n\n# 临时切换\nopenclaw models set yunyi-claude/claude-sonnet-4-20250514\n\n# 设置 fallback\nopenclaw models fallbacks add deepseek/deepseek-chat\n\n策略 2: Context 优化\n\n问题: 默认加载所有 context 文件（SOUL.md, AGENTS.md, TOOLS.md, MEMORY.md, docs/**/*.md），可能 50k+ tokens\n\n解决方案:\n\n精简 AGENTS.md\n\n移除冗余说明\n合并重复规则\n使用简洁语言\n\nLazy Loading\n\n简单任务只加载 SOUL.md + IDENTITY.md\n复杂任务按需加载相关文档\n参考 openclaw-token-optimizer 的 context_optimizer\n\n定期清理\n\n# 清理旧 memory logs\nfind ~/.openclaw/workspace/memory -name \"2026-*.md\" -mtime +30 -delete\n\n策略 3: Cron 优化\n\n识别高频 Cron:\n\nopenclaw cron list\n\n\n优化方案:\n\n非关键任务降低频率（每小时 → 每 4 小时）\n使用更便宜的模型执行 cron\n合并多个小任务为一个批处理\n\n示例:\n\n# 修改 cron 频率\nopenclaw cron edit \n\n# 为 cron 指定模型\nopenclaw cron add --model local/qwen2.5:7b \"0 */4 * * *\" \"检查系统状态\"\n\n策略 4: 会话管理\n\n长对话问题: Context 累积导致每次请求都携带完整历史\n\n解决方案:\n\n超过 50k tokens 时主动开启新会话\n使用 /new 命令清空 context\n重要信息保存到 MEMORY.md，不依赖会话历史\n策略 5: 启用本地模型\n\n适用场景:\n\n文件读取、简单查询\n开发测试\n离线工作\n\n设置:\n\n# 确保 Ollama 运行\nollama serve\n\n# 拉取模型\nollama pull qwen2.5:7b\n\n# 切换到本地模型\nopenclaw models set local/qwen2.5:7b\n\n成本基准\n\n基于实际使用数据的成本参考:\n\n使用模式\t每日会话数\t平均 tokens/会话\t每日成本\t每月成本\n轻度使用\t5-10\t20k\t$0.50-1.00\t$15-30\n中度使用\t20-30\t30k\t$2.00-4.00\t$60-120\n重度使用\t50+\t40k\t$8.00-15.00\t$240-450\n优化后\t50+\t15k\t$3.00-6.00\t$90-180\n\n优化目标: 重度使用场景下节省 50-60% 成本\n\n集成到工作流\n每日成本检查（推荐）\n\n添加 cron 任务，每天生成报告:\n\nopenclaw cron add \"0 9 * * *\" \"node ~/.openclaw/workspace/skills/openclaw-cost-optimizer/scripts/cost_analyzer.js quick\"\n\n每周深度分析\nopenclaw cron add \"0 10 * * 1\" \"node ~/.openclaw/workspace/skills/openclaw-cost-optimizer/scripts/cost_analyzer.js analyze 7\"\n\n成本告警\n\n在 AGENTS.md 中添加规则:\n\n## 成本监控\n- 每日成本超过 $5 → 立即告警\n- 单次会话超过 $1 → 记录并分析\n- 每周生成成本报告\n\n与其他 Skills 配合\nopenclaw-token-optimizer\n本 skill 专注成本分析和报告\ntoken-optimizer 提供底层优化工具（context_optimizer, model_router）\n配合使用效果最佳\nopenclaw-doctor\ndoctor 检查系统健康\ncost-optimizer 检查成本健康\n一起使用确保系统稳定且经济\n技术细节\n数据来源\nSession logs: ~/.openclaw/agents/main/agent/sessions/*.jsonl\n分析最近 N 天的日志（默认 7 天）\n提取 inputTokens, outputTokens, model, sessionId\n成本计算\n\n基于各模型官方定价:\n\nmodelCosts: {\n 'yunyi-claude/claude-opus-4-6': { input: 15, output: 75 },\n 'yunyi-claude/claude-sonnet-4-20250514': { input: 3, output: 15 },\n 'deepseek/deepseek-chat': { input: 0.014, output: 0.028 },\n 'local/qwen2.5:7b': { input: 0, output: 0 }\n}\n\n阈值配置\nthresholds: {\n longConversation: 50000, // tokens\n highContextSession: 30000, // tokens\n frequentCron: 10, // 每天次数\n expensiveModel: 10 // USD per 1M tokens\n}\n\n\n可在脚本中修改以适应个人需求。\n\n安全性\n\n✅ 纯本地运行\n\n无网络请求\n无外部依赖\n不执行子进程\n数据不离开本机\n\n✅ 只读分析\n\n只读取 session logs\n不修改任何配置\n不执行任何操作\n建议需手动执行\n故障排查\n问题: 未找到 session logs\n\n原因: logs 目录不存在或路径错误\n\n解决:\n\nls ~/.openclaw/agents/main/agent/sessions/\n\n\n如果目录不存在，说明还没有会话记录。\n\n问题: 成本计算不准确\n\n原因: 模型定价可能更新\n\n解决: 编辑 scripts/cost_analyzer.js，更新 modelCosts 对象。\n\n问题: 报告为空\n\n原因: 指定天数内没有日志\n\n解决: 增加分析天数\n\nnode scripts/cost_analyzer.js analyze 30\n\n更新日志\nv1.0.0 (2026-02-26)\n初始版本\n支持 session logs 分析\n识别 5 大高消耗场景\n生成优化建议和成本报告\n纯 Node.js 实现，无外部依赖\n贡献\n\n欢迎提交 Issue 和 PR:\n\n新的优化策略\n更准确的成本计算\n更多模型支持\n报告格式改进\n许可\n\nMIT License\n\n让 OpenClaw 更经济，让 AI 更可持续！ 💰✨" }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/dagangtj/oc-cost-analyzer", "publisherUrl": "https://clawhub.ai/dagangtj/oc-cost-analyzer", "owner": "dagangtj", "version": "1.0.1", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/oc-cost-analyzer", "downloadUrl": "https://openagent3.xyz/downloads/oc-cost-analyzer", "agentUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent", "manifestUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent.json", "briefUrl": "https://openagent3.xyz/skills/oc-cost-analyzer/agent.md" } }