From f288ddf444fdeb7659e7146ca25b16ea9a9bb882 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=BE=84=E6=BD=AD?= Date: Sun, 1 Feb 2026 12:23:15 +0800 Subject: [PATCH] feat(skill): add agent-session-monitor skill for LLM observability (#3426) --- .../agent-session-monitor/QUICKSTART.md | 138 ++++ .../skills/agent-session-monitor/README.md | 71 ++ .claude/skills/agent-session-monitor/SKILL.md | 384 +++++++++ .../example/clawdbot_demo.py | 101 +++ .../agent-session-monitor/example/demo.sh | 101 +++ .../agent-session-monitor/example/demo_v2.sh | 76 ++ .../example/test_access.log | 4 + .../example/test_access_v2.log | 4 + .../example/test_rotation.sh | 137 ++++ .claude/skills/agent-session-monitor/main.py | 639 +++++++++++++++ .../agent-session-monitor/scripts/cli.py | 600 ++++++++++++++ .../scripts/webserver.py | 755 ++++++++++++++++++ 12 files changed, 3010 insertions(+) create mode 100644 .claude/skills/agent-session-monitor/QUICKSTART.md create mode 100644 .claude/skills/agent-session-monitor/README.md create mode 100644 .claude/skills/agent-session-monitor/SKILL.md create mode 100755 .claude/skills/agent-session-monitor/example/clawdbot_demo.py create mode 100755 .claude/skills/agent-session-monitor/example/demo.sh create mode 100755 .claude/skills/agent-session-monitor/example/demo_v2.sh create mode 100644 .claude/skills/agent-session-monitor/example/test_access.log create mode 100644 .claude/skills/agent-session-monitor/example/test_access_v2.log create mode 100755 .claude/skills/agent-session-monitor/example/test_rotation.sh create mode 100755 .claude/skills/agent-session-monitor/main.py create mode 100755 .claude/skills/agent-session-monitor/scripts/cli.py create mode 100755 .claude/skills/agent-session-monitor/scripts/webserver.py diff --git a/.claude/skills/agent-session-monitor/QUICKSTART.md b/.claude/skills/agent-session-monitor/QUICKSTART.md new file mode 100644 index 000000000..fc72c7c1b --- /dev/null +++ b/.claude/skills/agent-session-monitor/QUICKSTART.md @@ -0,0 +1,138 @@ +# Agent Session Monitor - Quick Start + +实时Agent对话观测程序，用于监控Higress访问日志，追踪多轮对话的token开销和模型使用情况。 + +## 快速开始 + +### 1. 运行Demo + +```bash +cd example +bash demo.sh +``` + +这将： +- 解析示例日志文件 +- 列出所有session +- 显示session详细信息（包括完整的messages、question、answer、reasoning、tool_calls） +- 按模型和日期统计token开销 +- 导出FinOps报表 + +### 2. 启动Web界面（推荐） + +```bash +# 先解析日志生成session数据 +python3 main.py --log-path /var/log/higress/access.log --output-dir ./sessions + +# 启动Web服务器 +python3 scripts/webserver.py --data-dir ./sessions --port 8888 + +# 浏览器访问 +open http://localhost:8888 +``` + +Web界面功能： +- 📊 总览所有session，按模型分组统计 +- 🔍 点击session ID下钻查看完整对话 +- 💬 查看每轮的messages、question、answer、reasoning、tool_calls +- 💰 实时计算token开销和成本 +- 🔄 每30秒自动刷新 + +### 3. 在Clawdbot对话中使用 + +当用户询问当前会话token消耗时，生成观测链接： + +``` +你的当前会话ID: agent:main:discord:channel:1465367993012981988 + +查看详情：http://localhost:8888/session?id=agent:main:discord:channel:1465367993012981988 + +点击可以看到： +✅ 完整对话历史（每轮messages） +✅ Token消耗明细 +✅ 工具调用记录 +✅ 成本统计 +``` + +### 4. 使用CLI查询（可选） + +```bash +# 查看session详细信息 +python3 scripts/cli.py show + +# 列出所有session +python3 scripts/cli.py list + +# 按模型统计 +python3 scripts/cli.py stats-model + +# 导出报表 +python3 scripts/cli.py export finops-report.json +``` + +## 核心功能 + +✅ **完整对话追踪**：记录每轮对话的完整messages、question、answer、reasoning、tool_calls +✅ **Token开销统计**：区分input/output/reasoning/cached token，实时计算成本 +✅ **Session聚合**：按session_id关联多轮对话 +✅ **Web可视化界面**：浏览器访问，总览+下钻查看session详情 +✅ **实时URL生成**：Clawdbot可根据当前会话ID生成观测链接 +✅ **FinOps报表**：导出JSON/CSV格式的成本分析报告 + +## 日志格式要求 + +Higress访问日志需要包含ai_log字段（JSON格式），示例： + +```json +{ + "__file_offset__": "1000", + "timestamp": "2026-02-01T09:30:15Z", + "ai_log": "{\"session_id\":\"sess_abc\",\"messages\":[...],\"question\":\"...\",\"answer\":\"...\",\"input_token\":250,\"output_token\":160,\"model\":\"Qwen3-rerank\"}" +} +``` + +ai_log字段支持的属性： +- `session_id`: 会话标识（必需） +- `messages`: 完整对话历史 +- `question`: 当前轮次问题 +- `answer`: AI回答 +- `reasoning`: 思考过程（DeepSeek等模型） +- `tool_calls`: 工具调用列表 +- `input_token`: 输入token数 +- `output_token`: 输出token数 +- `model`: 模型名称 +- `response_type`: 响应类型 + +## 输出目录结构 + +``` +sessions/ +├── agent:main:discord:1465367993012981988.json +└── agent:test:discord:9999999999999999999.json +``` + +每个session文件包含： +- 基本信息（创建时间、更新时间、模型） +- Token统计（总输入、总输出、总reasoning、总cached） +- 对话轮次列表（每轮的完整messages、question、answer、reasoning、tool_calls） + +## 常见问题 + +**Q: 如何在Higress中配置session_id header？** +A: 在ai-statistics插件中配置`session_id_header`，或使用默认header（x-openclaw-session-key、x-clawdbot-session-key等）。详见PR #3420。 + +**Q: 支持哪些模型的pricing？** +A: 目前支持Qwen、DeepSeek、GPT-4、Claude等主流模型。可以在main.py的TOKEN_PRICING字典中添加新模型。 + +**Q: 如何实时监控日志文件变化？** +A: 直接运行main.py即可，程序使用定时轮询机制（每秒自动检查一次），无需安装额外依赖。 + +**Q: CLI查询速度慢？** +A: 大量session时，可以使用`--limit`限制结果数量，或按条件过滤（如`--sort-by cost`只查看成本最高的session）。 + +## 下一步 + +- 集成到Higress FinOps Dashboard +- 支持更多模型的pricing +- 添加趋势预测和异常检测 +- 支持多数据源聚合分析 diff --git a/.claude/skills/agent-session-monitor/README.md b/.claude/skills/agent-session-monitor/README.md new file mode 100644 index 000000000..c72a33d07 --- /dev/null +++ b/.claude/skills/agent-session-monitor/README.md @@ -0,0 +1,71 @@ +# Agent Session Monitor + +Real-time agent conversation monitoring for Clawdbot, designed to monitor Higress access logs and track token usage across multi-turn conversations. + +## Features + +- 🔍 **Complete Conversation Tracking**: Records messages, question, answer, reasoning, tool_calls for each turn +- 💰 **Token Usage Statistics**: Distinguishes input/output/reasoning/cached tokens, calculates costs in real-time +- 🌐 **Web Visualization**: Browser-based UI with overview and drill-down into session details +- 🔗 **Real-time URL Generation**: Clawdbot can generate observation links based on current session ID +- 🔄 **Log Rotation Support**: Automatically handles rotated log files (access.log, access.log.1, etc.) +- 📊 **FinOps Reporting**: Export usage data in JSON/CSV formats + +## Quick Start + +### 1. Run Demo + +```bash +cd example +bash demo.sh +``` + +### 2. Start Web UI + +```bash +# Parse logs +python3 main.py --log-path /var/log/higress/access.log --output-dir ./sessions + +# Start web server +python3 scripts/webserver.py --data-dir ./sessions --port 8888 + +# Access in browser +open http://localhost:8888 +``` + +### 3. Use in Clawdbot + +When users ask "How many tokens did this conversation use?", you can respond with: + +``` +Your current session statistics: +- Session ID: agent:main:discord:channel:1465367993012981988 +- View details: http://localhost:8888/session?id=agent:main:discord:channel:1465367993012981988 + +Click to see: +✅ Complete conversation history +✅ Token usage breakdown per turn +✅ Tool call records +✅ Cost statistics +``` + +## Files + +- `main.py`: Background monitor, parses Higress access logs +- `scripts/webserver.py`: Web server, provides browser-based UI +- `scripts/cli.py`: Command-line tools for queries and exports +- `example/`: Demo examples and test data + +## Dependencies + +- Python 3.8+ +- No external dependencies (uses only standard library) + +## Documentation + +- `SKILL.md`: Main skill documentation +- `QUICKSTART.md`: Quick start guide + +## License + +MIT diff --git a/.claude/skills/agent-session-monitor/SKILL.md b/.claude/skills/agent-session-monitor/SKILL.md new file mode 100644 index 000000000..1a4ee484f --- /dev/null +++ b/.claude/skills/agent-session-monitor/SKILL.md @@ -0,0 +1,384 @@ +--- +name: agent-session-monitor +description: Real-time agent conversation monitoring - monitors Higress access logs, aggregates conversations by session, tracks token usage. Supports web interface for viewing complete conversation history and costs. Use when users ask about current session token consumption, conversation history, or cost statistics. + +--- + +## Overview + +Real-time monitoring of Higress access logs, extracting ai_log JSON, grouping multi-turn conversations by session_id, and calculating token costs with visualization. + +### Core Features + +- **Real-time Log Monitoring**: Monitors Higress access log files, parses new ai_log entries in real-time +- **Log Rotation Support**: Full logrotate support, automatically tracks access.log.1~5 etc. +- **Incremental Parsing**: Inode-based tracking, processes only new content, no duplicates +- **Session Grouping**: Associates multi-turn conversations by session_id (each turn is a separate request) +- **Complete Conversation Tracking**: Records messages, question, answer, reasoning, tool_calls for each turn +- **Token Usage Tracking**: Distinguishes input/output/reasoning/cached tokens +- **Web Visualization**: Browser-based UI with overview and session drill-down +- **Real-time URL Generation**: Clawdbot can generate observation links based on current session ID +- **Background Processing**: Independent process, continuously parses access logs +- **State Persistence**: Maintains parsing progress and session data across runs + +## Usage + +### 1. Background Monitoring (Continuous) + +```bash +# Parse Higress access logs (with log rotation support) +python3 main.py --log-path /var/log/proxy/access.log --output-dir ./sessions + +# Filter by session key +python3 main.py --log-path /var/log/proxy/access.log --session-key + +# Scheduled task (incremental parsing every minute) +* * * * * python3 /path/to/main.py --log-path /var/log/proxy/access.log --output-dir /var/lib/sessions +``` + +**Log Rotation Notes**: +- Automatically scans `access.log`, `access.log.1`, `access.log.2`, etc. +- Uses inode tracking to identify files even after renaming +- State persistence prevents duplicate parsing +- Session data accumulates correctly across multiple runs + +See: [LOG_ROTATION.md](LOG_ROTATION.md) + +### 2. Start Web UI (Recommended) + +```bash +# Start web server +python3 scripts/webserver.py --data-dir ./sessions --port 8888 + +# Access in browser +open http://localhost:8888 +``` + +Web UI features: +- 📊 Overview: View all session statistics and group by model +- 🔍 Session Details: Click session ID to drill down into complete conversation history +- 💬 Conversation Log: Display messages, question, answer, reasoning, tool_calls for each turn +- 💰 Cost Statistics: Real-time token usage and cost calculation +- 🔄 Auto Refresh: Updates every 30 seconds + +### 3. Use in Clawdbot Conversations + +When users ask about current session token consumption or conversation history: + +1. Get current session_id (from runtime or context) +2. Generate web UI URL and return to user + +Example response: + +``` +Your current session statistics: +- Session ID: agent:main:discord:channel:1465367993012981988 +- View details: http://localhost:8888/session?id=agent:main:discord:channel:1465367993012981988 + +Click the link to see: +✅ Complete conversation history +✅ Token usage breakdown per turn +✅ Tool call records +✅ Cost statistics +``` + +### 4. CLI Queries (Optional) + +```bash +# View specific session details +python3 scripts/cli.py show + +# List all sessions +python3 scripts/cli.py list --sort-by cost --limit 10 + +# Statistics by model +python3 scripts/cli.py stats-model + +# Statistics by date (last 7 days) +python3 scripts/cli.py stats-date --days 7 + +# Export reports +python3 scripts/cli.py export finops-report.json +``` + +## Configuration + +### main.py (Background Monitor) + +| Parameter | Description | Required | Default | +|-----------|-------------|----------|---------| +| `--log-path` | Higress access log file path | Yes | /var/log/higress/access.log | +| `--output-dir` | Session data storage directory | No | ./sessions | +| `--session-key` | Monitor only specified session key | No | Monitor all sessions | +| `--state-file` | State file path (records read offsets) | No | /.state.json | +| `--refresh-interval` | Log refresh interval (seconds) | No | 1 | + +### webserver.py (Web UI) + +| Parameter | Description | Required | Default | +|-----------|-------------|----------|---------| +| `--data-dir` | Session data directory | No | ./sessions | +| `--port` | HTTP server port | No | 8888 | +| `--host` | HTTP server address | No | 0.0.0.0 | + +## Output Examples + +### 1. Real-time Monitor + +``` +🔍 Session Monitor - Active +━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ + +📊 Active Sessions: 3 + +┌──────────────────────────┬─────────┬──────────┬───────────┐ +│ Session ID │ Msgs │ Input │ Output │ +├──────────────────────────┼─────────┼──────────┼───────────┤ +│ sess_abc123 │ 5 │ 1,250 │ 800 │ +│ sess_xyz789 │ 3 │ 890 │ 650 │ +│ sess_def456 │ 8 │ 2,100 │ 1,200 │ +└──────────────────────────┴─────────┴──────────┴───────────┘ + +📈 Token Statistics + Total Input: 4240 tokens + Total Output: 2650 tokens + Total Cached: 0 tokens + Total Cost: $0.00127 +━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ +``` + +### 2. CLI Session Details + +```bash +$ python3 scripts/cli.py show agent:main:discord:channel:1465367993012981988 + +====================================================================== +📊 Session Detail: agent:main:discord:channel:1465367993012981988 +====================================================================== + +🕐 Created: 2026-02-01T09:30:00+08:00 +🕑 Updated: 2026-02-01T10:35:12+08:00 +🤖 Model: Qwen3-rerank +💬 Messages: 5 + +📈 Token Statistics: + Input: 1,250 tokens + Output: 800 tokens + Reasoning: 150 tokens + Total: 2,200 tokens + +💰 Estimated Cost: $0.00126000 USD + +📝 Conversation Rounds (5): +────────────────────────────────────────────────────────────────────── + + Round 1 @ 2026-02-01T09:30:15+08:00 + Tokens: 250 in → 160 out + 🔧 Tool calls: Yes + Messages (2): + [user] Check Beijing weather + ❓ Question: Check Beijing weather + ✅ Answer: Checking Beijing weather for you... + 🧠 Reasoning: User wants to know Beijing weather, I need to call weather API. + 🛠️ Tool Calls: + - get_weather({"location":"Beijing"}) +``` + +### 3. Statistics by Model + +```bash +$ python3 scripts/cli.py stats-model + +================================================================================ +📊 Statistics by Model +================================================================================ + +Model Sessions Input Output Cost (USD) +──────────────────────────────────────────────────────────────────────────── +Qwen3-rerank 12 15,230 9,840 $ 0.016800 +DeepSeek-R1 5 8,450 6,200 $ 0.010600 +Qwen-Max 3 4,200 3,100 $ 0.008300 +GPT-4 2 2,100 1,800 $ 0.017100 +──────────────────────────────────────────────────────────────────────────── +TOTAL 22 29,980 20,940 $ 0.052800 + +================================================================================ +``` + +### 4. Statistics by Date + +```bash +$ python3 scripts/cli.py stats-date --days 7 + +================================================================================ +📊 Statistics by Date (Last 7 days) +================================================================================ + +Date Sessions Input Output Cost (USD) Models +──────────────────────────────────────────────────────────────────────────── +2026-01-26 3 2,100 1,450 $ 0.0042 Qwen3-rerank +2026-01-27 5 4,850 3,200 $ 0.0096 Qwen3-rerank, GPT-4 +2026-01-28 4 3,600 2,800 $ 0.0078 DeepSeek-R1, Qwen +──────────────────────────────────────────────────────────────────────────── +TOTAL 22 29,980 20,940 $ 0.0528 + +================================================================================ +``` + +### 5. Web UI (Recommended) + +Access `http://localhost:8888` to see: + +**Home Page:** +- 📊 Total sessions, token consumption, cost cards +- 📋 Recent sessions list (clickable for details) +- 📈 Statistics by model table + +**Session Detail Page:** +- 💬 Complete conversation log (messages, question, answer, reasoning, tool_calls per turn) +- 🔧 Tool call history +- 💰 Token usage breakdown and costs + +**Features:** +- 🔄 Auto-refresh every 30 seconds +- 📱 Responsive design, mobile-friendly +- 🎨 Clean UI, easy to read + +## Session Data Structure + +Each session is stored as an independent JSON file with complete conversation history and token statistics: + +```json +{ + "session_id": "agent:main:discord:channel:1465367993012981988", + "created_at": "2026-02-01T10:30:00Z", + "updated_at": "2026-02-01T10:35:12Z", + "messages_count": 5, + "total_input_tokens": 1250, + "total_output_tokens": 800, + "total_reasoning_tokens": 150, + "total_cached_tokens": 0, + "model": "Qwen3-rerank", + "rounds": [ + { + "round": 1, + "timestamp": "2026-02-01T10:30:15Z", + "input_tokens": 250, + "output_tokens": 160, + "reasoning_tokens": 0, + "cached_tokens": 0, + "model": "Qwen3-rerank", + "has_tool_calls": true, + "response_type": "normal", + "messages": [ + { + "role": "system", + "content": "You are a helpful assistant..." + }, + { + "role": "user", + "content": "Check Beijing weather" + } + ], + "question": "Check Beijing weather", + "answer": "Checking Beijing weather for you...", + "reasoning": "User wants to know Beijing weather, need to call weather API.", + "tool_calls": [ + { + "index": 0, + "id": "call_abc123", + "type": "function", + "function": { + "name": "get_weather", + "arguments": "{\"location\":\"Beijing\"}" + } + } + ], + "input_token_details": {"cached_tokens": 0}, + "output_token_details": {} + } + ] +} +``` + +### Field Descriptions + +**Session Level:** +- `session_id`: Unique session identifier (from ai_log's session_id field) +- `created_at`: Session creation time +- `updated_at`: Last update time +- `messages_count`: Number of conversation turns +- `total_input_tokens`: Cumulative input tokens +- `total_output_tokens`: Cumulative output tokens +- `total_reasoning_tokens`: Cumulative reasoning tokens (DeepSeek, o1, etc.) +- `total_cached_tokens`: Cumulative cached tokens (prompt caching) +- `model`: Current model in use + +**Round Level (rounds):** +- `round`: Turn number +- `timestamp`: Current turn timestamp +- `input_tokens`: Input tokens for this turn +- `output_tokens`: Output tokens for this turn +- `reasoning_tokens`: Reasoning tokens (o1, etc.) +- `cached_tokens`: Cached tokens (prompt caching) +- `model`: Model used for this turn +- `has_tool_calls`: Whether includes tool calls +- `response_type`: Response type (normal/error, etc.) +- `messages`: Complete conversation history (OpenAI messages format) +- `question`: User's question for this turn (last user message) +- `answer`: AI's answer for this turn +- `reasoning`: AI's thinking process (if model supports) +- `tool_calls`: Tool call list (if any) +- `input_token_details`: Complete input token details (JSON) +- `output_token_details`: Complete output token details (JSON) + +## Log Format Requirements + +Higress access logs must include ai_log field (JSON format). Example: + +```json +{ + "__file_offset__": "1000", + "timestamp": "2026-02-01T09:30:15Z", + "ai_log": "{\"session_id\":\"sess_abc\",\"messages\":[...],\"question\":\"...\",\"answer\":\"...\",\"input_token\":250,\"output_token\":160,\"model\":\"Qwen3-rerank\"}" +} +``` + +Supported ai_log attributes: +- `session_id`: Session identifier (required) +- `messages`: Complete conversation history +- `question`: Question for current turn +- `answer`: AI answer +- `reasoning`: Thinking process (DeepSeek, o1, etc.) +- `reasoning_tokens`: Reasoning token count (from PR #3424) +- `cached_tokens`: Cached token count (from PR #3424) +- `tool_calls`: Tool call list +- `input_token`: Input token count +- `output_token`: Output token count +- `input_token_details`: Complete input token details (JSON) +- `output_token_details`: Complete output token details (JSON) +- `model`: Model name +- `response_type`: Response type + +## Implementation + +### Technology Stack + +- **Log Parsing**: Direct JSON parsing, no regex needed +- **File Monitoring**: Polling-based (no watchdog dependency) +- **Session Management**: In-memory + disk hybrid storage +- **Token Calculation**: Model-specific pricing for GPT-4, Qwen, Claude, o1, etc. + +### Privacy and Security + +- ✅ Does not record conversation content in logs, only token statistics +- ✅ Session data stored locally, not uploaded to external services +- ✅ Supports log file path allowlist +- ✅ Session key access control + +### Performance Optimization + +- Incremental log parsing, avoids full scans +- In-memory session data with periodic persistence +- Optimized log file reading (offset tracking) +- Inode-based file identification (handles rotation efficiently) diff --git a/.claude/skills/agent-session-monitor/example/clawdbot_demo.py b/.claude/skills/agent-session-monitor/example/clawdbot_demo.py new file mode 100755 index 000000000..9a7c9e564 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/clawdbot_demo.py @@ -0,0 +1,101 @@ +#!/usr/bin/env python3 +""" +演示如何在Clawdbot中生成Session观测URL +""" + +from urllib.parse import quote + +def generate_session_url(session_id: str, base_url: str = "http://localhost:8888") -> dict: + """ + 生成session观测URL + + Args: + session_id: 当前会话的session ID + base_url: Web服务器基础URL + + Returns: + 包含各种URL的字典 + """ + # URL编码session_id（处理特殊字符） + encoded_id = quote(session_id, safe='') + + return { + "session_detail": f"{base_url}/session?id={encoded_id}", + "api_session": f"{base_url}/api/session?id={encoded_id}", + "index": f"{base_url}/", + "api_sessions": f"{base_url}/api/sessions", + "api_stats": f"{base_url}/api/stats", + } + + +def format_response_message(session_id: str, base_url: str = "http://localhost:8888") -> str: + """ + 生成给用户的回复消息 + + Args: + session_id: 当前会话的session ID + base_url: Web服务器基础URL + + Returns: + 格式化的回复消息 + """ + urls = generate_session_url(session_id, base_url) + + return f"""你的当前会话信息： + +📊 **Session ID**: `{session_id}` + +🔗 **查看详情**: {urls['session_detail']} + +点击链接可以看到： +✅ 完整对话历史（每轮messages） +✅ Token消耗明细（input/output/reasoning） +✅ 工具调用记录 +✅ 实时成本统计 + +**更多链接：** +- 📋 所有会话: {urls['index']} +- 📥 API数据: {urls['api_session']} +- 📊 总体统计: {urls['api_stats']} +""" + + +# 示例使用 +if __name__ == '__main__': + # 模拟clawdbot的session ID + demo_session_id = "agent:main:discord:channel:1465367993012981988" + + print("=" * 70) + print("🤖 Clawdbot Session Monitor Demo") + print("=" * 70) + print() + + # 生成URL + urls = generate_session_url(demo_session_id) + + print("生成的URL：") + print(f" Session详情: {urls['session_detail']}") + print(f" API数据: {urls['api_session']}") + print(f" 总览页面: {urls['index']}") + print() + + # 生成回复消息 + message = format_response_message(demo_session_id) + + print("回复消息模板：") + print("-" * 70) + print(message) + print("-" * 70) + print() + + print("✅ 在Clawdbot中，你可以直接返回上面的消息给用户") + print() + + # 测试特殊字符的session ID + special_session_id = "agent:test:session/with?special&chars" + special_urls = generate_session_url(special_session_id) + + print("特殊字符处理示例：") + print(f" 原始ID: {special_session_id}") + print(f" URL: {special_urls['session_detail']}") + print() diff --git a/.claude/skills/agent-session-monitor/example/demo.sh b/.claude/skills/agent-session-monitor/example/demo.sh new file mode 100755 index 000000000..4a1e50d98 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/demo.sh @@ -0,0 +1,101 @@ +#!/bin/bash +# Agent Session Monitor - 演示脚本 + +set -e + +SKILL_DIR="$(dirname "$(dirname "$(realpath "$0")")")" +EXAMPLE_DIR="$SKILL_DIR/example" +LOG_FILE="$EXAMPLE_DIR/test_access.log" +OUTPUT_DIR="$EXAMPLE_DIR/sessions" + +echo "========================================" +echo "Agent Session Monitor - Demo" +echo "========================================" +echo "" + +# 清理旧数据 +if [ -d "$OUTPUT_DIR" ]; then + echo "🧹 Cleaning up old session data..." + rm -rf "$OUTPUT_DIR" +fi + +echo "📂 Log file: $LOG_FILE" +echo "📁 Output dir: $OUTPUT_DIR" +echo "" + +# 步骤1：解析日志文件（单次模式） +echo "========================================" +echo "步骤1：解析日志文件" +echo "========================================" +python3 "$SKILL_DIR/main.py" \ + --log-path "$LOG_FILE" \ + --output-dir "$OUTPUT_DIR" + +echo "" +echo "✅ 日志解析完成！Session数据已保存到: $OUTPUT_DIR" +echo "" + +# 步骤2：列出所有session +echo "========================================" +echo "步骤2：列出所有session" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" list \ + --data-dir "$OUTPUT_DIR" \ + --limit 10 + +# 步骤3：查看第一个session的详细信息 +echo "========================================" +echo "步骤3：查看session详细信息" +echo "========================================" +FIRST_SESSION=$(ls -1 "$OUTPUT_DIR"/*.json | head -1 | xargs -I {} basename {} .json) +python3 "$SKILL_DIR/scripts/cli.py" show "$FIRST_SESSION" \ + --data-dir "$OUTPUT_DIR" + +# 步骤4：按模型统计 +echo "========================================" +echo "步骤4：按模型统计token开销" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" stats-model \ + --data-dir "$OUTPUT_DIR" + +# 步骤5：按日期统计 +echo "========================================" +echo "步骤5：按日期统计token开销" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" stats-date \ + --data-dir "$OUTPUT_DIR" \ + --days 7 + +# 步骤6：导出FinOps报表 +echo "========================================" +echo "步骤6：导出FinOps报表" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" export "$EXAMPLE_DIR/finops-report.json" \ + --data-dir "$OUTPUT_DIR" \ + --format json + +echo "" +echo "✅ 报表已导出到: $EXAMPLE_DIR/finops-report.json" +echo "" + +# 显示报表内容 +if [ -f "$EXAMPLE_DIR/finops-report.json" ]; then + echo "📊 FinOps报表内容：" + echo "========================================" + cat "$EXAMPLE_DIR/finops-report.json" | python3 -m json.tool | head -50 + echo "..." +fi + +echo "" +echo "========================================" +echo "✅ Demo完成！" +echo "========================================" +echo "" +echo "💡 提示：" +echo " - Session数据保存在: $OUTPUT_DIR/" +echo " - FinOps报表: $EXAMPLE_DIR/finops-report.json" +echo " - 使用 'python3 scripts/cli.py --help' 查看更多命令" +echo "" +echo "🌐 启动Web界面查看：" +echo " python3 $SKILL_DIR/scripts/webserver.py --data-dir $OUTPUT_DIR --port 8888" +echo " 然后访问: http://localhost:8888" diff --git a/.claude/skills/agent-session-monitor/example/demo_v2.sh b/.claude/skills/agent-session-monitor/example/demo_v2.sh new file mode 100755 index 000000000..fa311e8e4 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/demo_v2.sh @@ -0,0 +1,76 @@ +#!/bin/bash +# Agent Session Monitor - Demo for PR #3424 token details + +set -e + +SKILL_DIR="$(dirname "$(dirname "$(realpath "$0")")")" +EXAMPLE_DIR="$SKILL_DIR/example" +LOG_FILE="$EXAMPLE_DIR/test_access_v2.log" +OUTPUT_DIR="$EXAMPLE_DIR/sessions_v2" + +echo "========================================" +echo "Agent Session Monitor - Token Details Demo" +echo "========================================" +echo "" + +# 清理旧数据 +if [ -d "$OUTPUT_DIR" ]; then + echo "🧹 Cleaning up old session data..." + rm -rf "$OUTPUT_DIR" +fi + +echo "📂 Log file: $LOG_FILE" +echo "📁 Output dir: $OUTPUT_DIR" +echo "" + +# 步骤1：解析日志文件 +echo "========================================" +echo "步骤1：解析日志文件（包含token details）" +echo "========================================" +python3 "$SKILL_DIR/main.py" \ + --log-path "$LOG_FILE" \ + --output-dir "$OUTPUT_DIR" + +echo "" +echo "✅ 日志解析完成！Session数据已保存到: $OUTPUT_DIR" +echo "" + +# 步骤2：查看使用prompt caching的session（gpt-4o） +echo "========================================" +echo "步骤2：查看GPT-4o session（包含cached tokens）" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" show "agent:main:discord:1465367993012981988" \ + --data-dir "$OUTPUT_DIR" + +# 步骤3：查看使用reasoning的session（o1） +echo "========================================" +echo "步骤3：查看o1 session（包含reasoning tokens）" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" show "agent:main:discord:9999999999999999999" \ + --data-dir "$OUTPUT_DIR" + +# 步骤4：按模型统计 +echo "========================================" +echo "步骤4：按模型统计（包含新token类型）" +echo "========================================" +python3 "$SKILL_DIR/scripts/cli.py" stats-model \ + --data-dir "$OUTPUT_DIR" + +echo "" +echo "========================================" +echo "✅ Demo完成！" +echo "========================================" +echo "" +echo "💡 新功能说明：" +echo " ✅ cached_tokens - 缓存命中的token数（prompt caching）" +echo " ✅ reasoning_tokens - 推理token数（o1等模型）" +echo " ✅ input_token_details - 完整输入token详情（JSON）" +echo " ✅ output_token_details - 完整输出token详情（JSON）" +echo "" +echo "💰 成本计算已优化：" +echo " - cached tokens通常比regular input便宜（50-90%折扣）" +echo " - reasoning tokens单独计费（o1系列）" +echo "" +echo "🌐 启动Web界面查看：" +echo " python3 $SKILL_DIR/scripts/webserver.py --data-dir $OUTPUT_DIR --port 8889" +echo " 然后访问: http://localhost:8889" diff --git a/.claude/skills/agent-session-monitor/example/test_access.log b/.claude/skills/agent-session-monitor/example/test_access.log new file mode 100644 index 000000000..34f3aa4f2 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/test_access.log @@ -0,0 +1,4 @@ +{"__file_offset__":"1000","timestamp":"2026-02-01T09:30:15Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"Qwen3-rerank@higress\",\"api_type\":\"LLM\",\"chat_round\":1,\"consumer\":\"clawdbot\",\"input_token\":250,\"output_token\":160,\"model\":\"Qwen3-rerank\",\"response_type\":\"normal\",\"total_token\":410,\"messages\":[{\"role\":\"system\",\"content\":\"You are a helpful assistant.\"},{\"role\":\"user\",\"content\":\"查询北京天气\"}],\"question\":\"查询北京天气\",\"answer\":\"正在为您查询北京天气...\",\"reasoning\":\"用户想知道北京的天气，我需要调用天气查询工具。\",\"tool_calls\":[{\"index\":0,\"id\":\"call_abc123\",\"type\":\"function\",\"function\":{\"name\":\"get_weather\",\"arguments\":\"{\\\"location\\\":\\\"Beijing\\\"}\"}}]}"} +{"__file_offset__":"2000","timestamp":"2026-02-01T09:32:00Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"Qwen3-rerank@higress\",\"api_type\":\"LLM\",\"chat_round\":2,\"consumer\":\"clawdbot\",\"input_token\":320,\"output_token\":180,\"model\":\"Qwen3-rerank\",\"response_type\":\"normal\",\"total_token\":500,\"messages\":[{\"role\":\"tool\",\"content\":\"{\\\"temperature\\\": 15, \\\"weather\\\": \\\"晴\\\"}\"}],\"question\":\"\",\"answer\":\"北京今天天气晴朗，温度15°C。\",\"reasoning\":\"\",\"tool_calls\":[]}"} +{"__file_offset__":"3000","timestamp":"2026-02-01T09:35:12Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"Qwen3-rerank@higress\",\"api_type\":\"LLM\",\"chat_round\":3,\"consumer\":\"clawdbot\",\"input_token\":380,\"output_token\":220,\"model\":\"Qwen3-rerank\",\"response_type\":\"normal\",\"total_token\":600,\"messages\":[{\"role\":\"user\",\"content\":\"谢谢！\"},{\"role\":\"assistant\",\"content\":\"不客气！如果还有其他问题，随时问我。\"}],\"question\":\"谢谢！\",\"answer\":\"不客气！如果还有其他问题，随时问我。\",\"reasoning\":\"\",\"tool_calls\":[]}"} +{"__file_offset__":"4000","timestamp":"2026-02-01T10:00:00Z","ai_log":"{\"session_id\":\"agent:test:discord:9999999999999999999\",\"api\":\"DeepSeek-R1@higress\",\"api_type\":\"LLM\",\"chat_round\":1,\"consumer\":\"clawdbot\",\"input_token\":50,\"output_token\":30,\"model\":\"DeepSeek-R1\",\"response_type\":\"normal\",\"total_token\":80,\"messages\":[{\"role\":\"user\",\"content\":\"计算2+2\"}],\"question\":\"计算2+2\",\"answer\":\"4\",\"reasoning\":\"这是一个简单的加法运算，2加2等于4。\",\"tool_calls\":[]}"} diff --git a/.claude/skills/agent-session-monitor/example/test_access_v2.log b/.claude/skills/agent-session-monitor/example/test_access_v2.log new file mode 100644 index 000000000..5ce055e40 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/test_access_v2.log @@ -0,0 +1,4 @@ +{"__file_offset__":"1000","timestamp":"2026-02-01T10:00:00Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"gpt-4o\",\"api_type\":\"LLM\",\"chat_round\":1,\"consumer\":\"clawdbot\",\"input_token\":150,\"output_token\":100,\"reasoning_tokens\":0,\"cached_tokens\":120,\"input_token_details\":\"{\\\"cached_tokens\\\":120}\",\"output_token_details\":\"{}\",\"model\":\"gpt-4o\",\"response_type\":\"normal\",\"total_token\":250,\"messages\":[{\"role\":\"system\",\"content\":\"You are a helpful assistant.\"},{\"role\":\"user\",\"content\":\"你好\"}],\"question\":\"你好\",\"answer\":\"你好！有什么我可以帮助你的吗？\",\"reasoning\":\"\",\"tool_calls\":[]}"} +{"__file_offset__":"2000","timestamp":"2026-02-01T10:01:00Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"gpt-4o\",\"api_type\":\"LLM\",\"chat_round\":2,\"consumer\":\"clawdbot\",\"input_token\":200,\"output_token\":150,\"reasoning_tokens\":0,\"cached_tokens\":80,\"input_token_details\":\"{\\\"cached_tokens\\\":80}\",\"output_token_details\":\"{}\",\"model\":\"gpt-4o\",\"response_type\":\"normal\",\"total_token\":350,\"messages\":[{\"role\":\"user\",\"content\":\"介绍一下你的能力\"}],\"question\":\"介绍一下你的能力\",\"answer\":\"我可以帮助你回答问题、写作、编程等...\",\"reasoning\":\"\",\"tool_calls\":[]}"} +{"__file_offset__":"3000","timestamp":"2026-02-01T10:02:00Z","ai_log":"{\"session_id\":\"agent:main:discord:9999999999999999999\",\"api\":\"o1\",\"api_type\":\"LLM\",\"chat_round\":1,\"consumer\":\"clawdbot\",\"input_token\":100,\"output_token\":80,\"reasoning_tokens\":500,\"cached_tokens\":0,\"input_token_details\":\"{}\",\"output_token_details\":\"{\\\"reasoning_tokens\\\":500}\",\"model\":\"o1\",\"response_type\":\"normal\",\"total_token\":580,\"messages\":[{\"role\":\"user\",\"content\":\"解释量子纠缠\"}],\"question\":\"解释量子纠缠\",\"answer\":\"量子纠缠是量子力学中的一种现象...\",\"reasoning\":\"这是一个复杂的物理概念，我需要仔细思考如何用简单的方式解释...\",\"tool_calls\":[]}"} +{"__file_offset__":"4000","timestamp":"2026-02-01T10:03:00Z","ai_log":"{\"session_id\":\"agent:main:discord:1465367993012981988\",\"api\":\"gpt-4o\",\"api_type\":\"LLM\",\"chat_round\":3,\"consumer\":\"clawdbot\",\"input_token\":300,\"output_token\":200,\"reasoning_tokens\":0,\"cached_tokens\":200,\"input_token_details\":\"{\\\"cached_tokens\\\":200}\",\"output_token_details\":\"{}\",\"model\":\"gpt-4o\",\"response_type\":\"normal\",\"total_token\":500,\"messages\":[{\"role\":\"user\",\"content\":\"写一个Python函数计算斐波那契数列\"}],\"question\":\"写一个Python函数计算斐波那契数列\",\"answer\":\"```python\\ndef fibonacci(n):\\n if n <= 1:\\n return n\\n return fibonacci(n-1) + fibonacci(n-2)\\n```\",\"reasoning\":\"\",\"tool_calls\":[]}"} diff --git a/.claude/skills/agent-session-monitor/example/test_rotation.sh b/.claude/skills/agent-session-monitor/example/test_rotation.sh new file mode 100755 index 000000000..076970b42 --- /dev/null +++ b/.claude/skills/agent-session-monitor/example/test_rotation.sh @@ -0,0 +1,137 @@ +#!/bin/bash +# 测试日志轮转功能 + +set -e + +SKILL_DIR="$(dirname "$(dirname "$(realpath "$0")")")" +EXAMPLE_DIR="$SKILL_DIR/example" +TEST_DIR="$EXAMPLE_DIR/rotation_test" +LOG_FILE="$TEST_DIR/access.log" +OUTPUT_DIR="$TEST_DIR/sessions" + +echo "========================================" +echo "Log Rotation Test" +echo "========================================" +echo "" + +# 清理旧测试数据 +rm -rf "$TEST_DIR" +mkdir -p "$TEST_DIR" + +echo "📁 Test directory: $TEST_DIR" +echo "" + +# 模拟日志轮转场景 +echo "========================================" +echo "步骤1：创建初始日志文件" +echo "========================================" + +# 创建第一批日志（10条） +for i in {1..10}; do + echo "{\"timestamp\":\"2026-02-01T10:0${i}:00Z\",\"ai_log\":\"{\\\"session_id\\\":\\\"session_001\\\",\\\"model\\\":\\\"gpt-4o\\\",\\\"input_token\\\":$((100+i)),\\\"output_token\\\":$((50+i)),\\\"cached_tokens\\\":$((30+i))}\"}" >> "$LOG_FILE" +done + +echo "✅ Created $LOG_FILE with 10 lines" +echo "" + +# 首次解析 +echo "========================================" +echo "步骤2：首次解析（应该处理10条记录）" +echo "========================================" +python3 "$SKILL_DIR/main.py" \ + --log-path "$LOG_FILE" \ + --output-dir "$OUTPUT_DIR" \ + + +echo "" + +# 检查session数据 +echo "Session数据：" +cat "$OUTPUT_DIR/session_001.json" | python3 -c "import sys, json; d=json.load(sys.stdin); print(f\" Messages: {d['messages_count']}, Total Input: {d['total_input_tokens']}\")" +echo "" + +# 模拟日志轮转 +echo "========================================" +echo "步骤3：模拟日志轮转" +echo "========================================" +mv "$LOG_FILE" "$LOG_FILE.1" +echo "✅ Rotated: access.log -> access.log.1" +echo "" + +# 创建新的日志文件（5条新记录） +for i in {11..15}; do + echo "{\"timestamp\":\"2026-02-01T10:${i}:00Z\",\"ai_log\":\"{\\\"session_id\\\":\\\"session_001\\\",\\\"model\\\":\\\"gpt-4o\\\",\\\"input_token\\\":$((100+i)),\\\"output_token\\\":$((50+i)),\\\"cached_tokens\\\":$((30+i))}\"}" >> "$LOG_FILE" +done + +echo "✅ Created new $LOG_FILE with 5 lines" +echo "" + +# 再次解析（应该只处理新的5条） +echo "========================================" +echo "步骤4：再次解析（应该只处理新的5条）" +echo "========================================" +python3 "$SKILL_DIR/main.py" \ + --log-path "$LOG_FILE" \ + --output-dir "$OUTPUT_DIR" \ + + +echo "" + +# 检查session数据 +echo "Session数据：" +cat "$OUTPUT_DIR/session_001.json" | python3 -c "import sys, json; d=json.load(sys.stdin); print(f\" Messages: {d['messages_count']}, Total Input: {d['total_input_tokens']} (应该是15条记录)\")" +echo "" + +# 再次轮转 +echo "========================================" +echo "步骤5：再次轮转" +echo "========================================" +mv "$LOG_FILE.1" "$LOG_FILE.2" +mv "$LOG_FILE" "$LOG_FILE.1" +echo "✅ Rotated: access.log -> access.log.1" +echo "✅ Rotated: access.log.1 -> access.log.2" +echo "" + +# 创建新的日志文件（3条新记录） +for i in {16..18}; do + echo "{\"timestamp\":\"2026-02-01T10:${i}:00Z\",\"ai_log\":\"{\\\"session_id\\\":\\\"session_001\\\",\\\"model\\\":\\\"gpt-4o\\\",\\\"input_token\\\":$((100+i)),\\\"output_token\\\":$((50+i)),\\\"cached_tokens\\\":$((30+i))}\"}" >> "$LOG_FILE" +done + +echo "✅ Created new $LOG_FILE with 3 lines" +echo "" + +# 再次解析（应该只处理新的3条） +echo "========================================" +echo "步骤6：再次解析（应该只处理新的3条）" +echo "========================================" +python3 "$SKILL_DIR/main.py" \ + --log-path "$LOG_FILE" \ + --output-dir "$OUTPUT_DIR" \ + + +echo "" + +# 检查session数据 +echo "Session数据：" +cat "$OUTPUT_DIR/session_001.json" | python3 -c "import sys, json; d=json.load(sys.stdin); print(f\" Messages: {d['messages_count']}, Total Input: {d['total_input_tokens']} (应该是18条记录)\")" +echo "" + +# 检查状态文件 +echo "========================================" +echo "步骤7：查看状态文件" +echo "========================================" +echo "状态文件内容：" +cat "$OUTPUT_DIR/.state.json" | python3 -m json.tool | head -20 +echo "" + +echo "========================================" +echo "✅ 测试完成！" +echo "========================================" +echo "" +echo "💡 验证要点：" +echo " 1. 首次解析处理了10条记录" +echo " 2. 轮转后只处理新增的5条记录（总计15条）" +echo " 3. 再次轮转后只处理新增的3条记录（总计18条）" +echo " 4. 状态文件记录了每个文件的inode和offset" +echo "" +echo "📂 测试数据保存在: $TEST_DIR/" diff --git a/.claude/skills/agent-session-monitor/main.py b/.claude/skills/agent-session-monitor/main.py new file mode 100755 index 000000000..5de348ac9 --- /dev/null +++ b/.claude/skills/agent-session-monitor/main.py @@ -0,0 +1,639 @@ +#!/usr/bin/env python3 +""" +Agent Session Monitor - 实时Agent对话观测程序 +监控Higress访问日志，按session聚合对话，追踪token开销 +""" + +import argparse +import json +import re +import os +import sys +import time +from collections import defaultdict +from datetime import datetime +from pathlib import Path +from typing import Dict, List, Optional + +# 使用定时轮询机制，不依赖watchdog + +# ============================================================================ +# 配置 +# ============================================================================ + +# Token定价（单位：美元/1M tokens） +TOKEN_PRICING = { + "Qwen": { + "input": 0.0002, # $0.2/1M + "output": 0.0006, + "cached": 0.0001, # cached tokens通常是input的50% + }, + "Qwen3-rerank": { + "input": 0.0003, + "output": 0.0012, + "cached": 0.00015, + }, + "Qwen-Max": { + "input": 0.0005, + "output": 0.002, + "cached": 0.00025, + }, + "GPT-4": { + "input": 0.003, + "output": 0.006, + "cached": 0.0015, + }, + "GPT-4o": { + "input": 0.0025, + "output": 0.01, + "cached": 0.00125, # GPT-4o prompt caching: 50% discount + }, + "GPT-4-32k": { + "input": 0.01, + "output": 0.03, + "cached": 0.005, + }, + "o1": { + "input": 0.015, + "output": 0.06, + "cached": 0.0075, + "reasoning": 0.06, # o1 reasoning tokens same as output + }, + "o1-mini": { + "input": 0.003, + "output": 0.012, + "cached": 0.0015, + "reasoning": 0.012, + }, + "Claude": { + "input": 0.015, + "output": 0.075, + "cached": 0.0015, # Claude prompt caching: 90% discount + }, + "DeepSeek-R1": { + "input": 0.004, + "output": 0.012, + "reasoning": 0.002, + "cached": 0.002, + } +} + +DEFAULT_LOG_PATH = "/var/log/higress/access.log" +DEFAULT_OUTPUT_DIR = "./sessions" + +# ============================================================================ +# Session管理器 +# ============================================================================ + +class SessionManager: + """管理多个会话的token统计""" + + def __init__(self, output_dir: str, load_existing: bool = True): + self.output_dir = Path(output_dir) + self.output_dir.mkdir(parents=True, exist_ok=True) + self.sessions: Dict[str, dict] = {} + + # 加载已有的session数据 + if load_existing: + self._load_existing_sessions() + + def _load_existing_sessions(self): + """加载已有的session数据""" + loaded_count = 0 + for session_file in self.output_dir.glob("*.json"): + try: + with open(session_file, 'r', encoding='utf-8') as f: + session = json.load(f) + session_id = session.get('session_id') + if session_id: + self.sessions[session_id] = session + loaded_count += 1 + except Exception as e: + print(f"Warning: Failed to load session {session_file}: {e}", file=sys.stderr) + + if loaded_count > 0: + print(f"📦 Loaded {loaded_count} existing session(s)") + + def update_session(self, session_id: str, ai_log: dict) -> dict: + """更新或创建session""" + if session_id not in self.sessions: + self.sessions[session_id] = { + "session_id": session_id, + "created_at": datetime.now().isoformat(), + "updated_at": datetime.now().isoformat(), + "messages_count": 0, + "total_input_tokens": 0, + "total_output_tokens": 0, + "total_reasoning_tokens": 0, + "total_cached_tokens": 0, + "rounds": [], + "model": ai_log.get("model", "unknown") + } + + session = self.sessions[session_id] + + # 更新统计 + model = ai_log.get("model", "unknown") + session["model"] = model + session["updated_at"] = datetime.now().isoformat() + + # Token统计 + session["total_input_tokens"] += ai_log.get("input_token", 0) + session["total_output_tokens"] += ai_log.get("output_token", 0) + + # 检查reasoning tokens（优先使用ai_log中的reasoning_tokens字段） + reasoning_tokens = ai_log.get("reasoning_tokens", 0) + if reasoning_tokens == 0 and "reasoning" in ai_log and ai_log["reasoning"]: + # 如果没有reasoning_tokens字段，估算reasoning的token数（大致按字符数/4） + reasoning_text = ai_log["reasoning"] + reasoning_tokens = len(reasoning_text) // 4 + session["total_reasoning_tokens"] += reasoning_tokens + + # 检查cached tokens（prompt caching） + cached_tokens = ai_log.get("cached_tokens", 0) + session["total_cached_tokens"] += cached_tokens + + # 检查是否有tool_calls（工具调用） + has_tool_calls = "tool_calls" in ai_log and ai_log["tool_calls"] + + # 更新消息数 + session["messages_count"] += 1 + + # 解析token details（如果有） + input_token_details = {} + output_token_details = {} + + if "input_token_details" in ai_log: + try: + # input_token_details可能是字符串或字典 + details = ai_log["input_token_details"] + if isinstance(details, str): + import json + input_token_details = json.loads(details) + else: + input_token_details = details + except (json.JSONDecodeError, TypeError): + pass + + if "output_token_details" in ai_log: + try: + # output_token_details可能是字符串或字典 + details = ai_log["output_token_details"] + if isinstance(details, str): + import json + output_token_details = json.loads(details) + else: + output_token_details = details + except (json.JSONDecodeError, TypeError): + pass + + # 添加轮次记录（包含完整的llm请求和响应信息） + round_data = { + "round": session["messages_count"], + "timestamp": datetime.now().isoformat(), + "input_tokens": ai_log.get("input_token", 0), + "output_tokens": ai_log.get("output_token", 0), + "reasoning_tokens": reasoning_tokens, + "cached_tokens": cached_tokens, + "model": model, + "has_tool_calls": has_tool_calls, + "response_type": ai_log.get("response_type", "normal"), + # 完整的对话信息 + "messages": ai_log.get("messages", []), + "question": ai_log.get("question", ""), + "answer": ai_log.get("answer", ""), + "reasoning": ai_log.get("reasoning", ""), + "tool_calls": ai_log.get("tool_calls", []), + # Token详情 + "input_token_details": input_token_details, + "output_token_details": output_token_details, + } + session["rounds"].append(round_data) + + # 保存到文件 + self._save_session(session) + + return session + + def _save_session(self, session: dict): + """保存session数据到文件""" + session_file = self.output_dir / f"{session['session_id']}.json" + with open(session_file, 'w', encoding='utf-8') as f: + json.dump(session, f, ensure_ascii=False, indent=2) + + def get_all_sessions(self) -> List[dict]: + """获取所有session""" + return list(self.sessions.values()) + + def get_session(self, session_id: str) -> Optional[dict]: + """获取指定session""" + return self.sessions.get(session_id) + + def get_summary(self) -> dict: + """获取总体统计""" + total_input = sum(s["total_input_tokens"] for s in self.sessions.values()) + total_output = sum(s["total_output_tokens"] for s in self.sessions.values()) + total_reasoning = sum(s.get("total_reasoning_tokens", 0) for s in self.sessions.values()) + total_cached = sum(s.get("total_cached_tokens", 0) for s in self.sessions.values()) + + # 计算成本 + total_cost = 0 + for session in self.sessions.values(): + model = session.get("model", "unknown") + input_tokens = session["total_input_tokens"] + output_tokens = session["total_output_tokens"] + reasoning_tokens = session.get("total_reasoning_tokens", 0) + cached_tokens = session.get("total_cached_tokens", 0) + + pricing = TOKEN_PRICING.get(model, TOKEN_PRICING.get("GPT-4", {})) + + # 基础成本计算 + # 注意：cached_tokens已经包含在input_tokens中，需要分开计算 + regular_input_tokens = input_tokens - cached_tokens + input_cost = regular_input_tokens * pricing.get("input", 0) / 1000000 + output_cost = output_tokens * pricing.get("output", 0) / 1000000 + + # reasoning成本 + reasoning_cost = 0 + if "reasoning" in pricing and reasoning_tokens > 0: + reasoning_cost = reasoning_tokens * pricing["reasoning"] / 1000000 + + # cached成本（通常比input便宜） + cached_cost = 0 + if "cached" in pricing and cached_tokens > 0: + cached_cost = cached_tokens * pricing["cached"] / 1000000 + + total_cost += input_cost + output_cost + reasoning_cost + cached_cost + + return { + "total_sessions": len(self.sessions), + "total_input_tokens": total_input, + "total_output_tokens": total_output, + "total_reasoning_tokens": total_reasoning, + "total_cached_tokens": total_cached, + "total_tokens": total_input + total_output + total_reasoning + total_cached, + "total_cost_usd": round(total_cost, 4), + "active_session_ids": list(self.sessions.keys()) + } + + +# ============================================================================ +# 日志解析器 +# ============================================================================ + +class LogParser: + """解析Higress访问日志，提取ai_log，支持日志轮转""" + + def __init__(self, state_file: str = None): + self.state_file = Path(state_file) if state_file else None + self.file_offsets = {} # {文件路径: 已读取的字节偏移} + self._load_state() + + def _load_state(self): + """加载上次的读取状态""" + if self.state_file and self.state_file.exists(): + try: + with open(self.state_file, 'r') as f: + self.file_offsets = json.load(f) + except Exception as e: + print(f"Warning: Failed to load state file: {e}", file=sys.stderr) + + def _save_state(self): + """保存当前的读取状态""" + if self.state_file: + try: + self.state_file.parent.mkdir(parents=True, exist_ok=True) + with open(self.state_file, 'w') as f: + json.dump(self.file_offsets, f, indent=2) + except Exception as e: + print(f"Warning: Failed to save state file: {e}", file=sys.stderr) + + def parse_log_line(self, line: str) -> Optional[dict]: + """解析单行日志，提取ai_log JSON""" + try: + # 直接解析整个日志行为JSON + log_obj = json.loads(line.strip()) + + # 获取ai_log字段（这是一个JSON字符串） + if 'ai_log' in log_obj: + ai_log_str = log_obj['ai_log'] + + # 解析内层JSON + ai_log = json.loads(ai_log_str) + return ai_log + except (json.JSONDecodeError, ValueError, KeyError): + # 静默忽略非JSON行或缺少ai_log字段的行 + pass + + return None + + def parse_rotated_logs(self, log_pattern: str, session_manager) -> None: + """解析日志文件及其轮转文件 + + Args: + log_pattern: 日志文件路径，如 /var/log/proxy/access.log + session_manager: Session管理器 + """ + base_path = Path(log_pattern) + + # 自动扫描所有轮转的日志文件（从旧到新） + log_files = [] + + # 自动扫描轮转文件（最多扫描到 .100，超过这个数量的日志应该很少见） + for i in range(100, 0, -1): + rotated_path = Path(f"{log_pattern}.{i}") + if rotated_path.exists(): + log_files.append(str(rotated_path)) + + # 添加当前日志文件 + if base_path.exists(): + log_files.append(str(base_path)) + + if not log_files: + print(f"❌ No log files found for pattern: {log_pattern}") + return + + print(f"📂 Found {len(log_files)} log file(s):") + for f in log_files: + print(f" - {f}") + print() + + # 按顺序解析每个文件（从旧到新） + for log_file in log_files: + self._parse_file_incremental(log_file, session_manager) + + # 保存状态 + self._save_state() + + def _parse_file_incremental(self, file_path: str, session_manager) -> None: + """增量解析单个日志文件""" + try: + file_stat = os.stat(file_path) + file_size = file_stat.st_size + file_inode = file_stat.st_ino + + # 使用inode作为主键 + inode_key = str(file_inode) + last_offset = self.file_offsets.get(inode_key, 0) + + # 如果文件变小了，说明是新文件（被truncate或新创建），从头开始读 + if file_size < last_offset: + print(f" 📝 File truncated or recreated, reading from start: {file_path}") + last_offset = 0 + + # 如果offset相同，说明没有新内容 + if file_size == last_offset: + print(f" ⏭️ No new content in: {file_path} (inode:{inode_key})") + return + + print(f" 📖 Reading {file_path} from offset {last_offset} to {file_size} (inode:{inode_key})") + + with open(file_path, 'r', encoding='utf-8', errors='ignore') as f: + f.seek(last_offset) + lines_processed = 0 + + for line in f: + ai_log = self.parse_log_line(line) + if ai_log: + session_id = ai_log.get("session_id", "default") + session_manager.update_session(session_id, ai_log) + lines_processed += 1 + + # 每处理1000行打印一次进度 + if lines_processed % 1000 == 0: + print(f" Processed {lines_processed} lines, {len(session_manager.sessions)} sessions") + + # 更新offset（使用inode作为key） + current_offset = f.tell() + self.file_offsets[inode_key] = current_offset + + print(f" ✅ Processed {lines_processed} new lines from {file_path}") + + except FileNotFoundError: + print(f" ❌ File not found: {file_path}") + except Exception as e: + print(f" ❌ Error parsing {file_path}: {e}") + + +# ============================================================================ +# 实时显示器 +# ============================================================================ + +class RealtimeMonitor: + """实时监控显示和交互（定时轮询模式）""" + + def __init__(self, session_manager: SessionManager, log_parser=None, log_path: str = None, refresh_interval: int = 1): + self.session_manager = session_manager + self.log_parser = log_parser + self.log_path = log_path + self.refresh_interval = refresh_interval + self.running = True + self.last_poll_time = 0 + + def start(self): + """启动实时监控（定时轮询日志文件）""" + print(f"\n{'=' * 50}") + print(f"🔍 Agent Session Monitor - Real-time View") + print(f"{'=' * 50}") + print() + print("Press Ctrl+C to stop...") + print() + + try: + while self.running: + # 定时轮询日志文件（检查新增内容和轮转） + current_time = time.time() + if self.log_parser and self.log_path and (current_time - self.last_poll_time >= self.refresh_interval): + self.log_parser.parse_rotated_logs(self.log_path, self.session_manager) + self.last_poll_time = current_time + + # 显示状态 + self._display_status() + time.sleep(self.refresh_interval) + except KeyboardInterrupt: + print("\n\n👋 Stopping monitor...") + self.running = False + self._display_summary() + + def _display_status(self): + """显示当前状态""" + summary = self.session_manager.get_summary() + + # 清屏 + os.system('clear' if os.name == 'posix' else 'cls') + + print(f"{'=' * 50}") + print(f"🔍 Session Monitor - Active") + print(f"{'=' * 50}") + print() + print(f"📊 Active Sessions: {summary['total_sessions']}") + print() + + # 显示活跃session的token统计 + if summary['active_session_ids']: + print("┌──────────────────────────┬─────────┬──────────┬───────────┐") + print("│ Session ID │ Msgs │ Input │ Output │") + print("├──────────────────────────┼─────────┼──────────┼───────────┤") + + for session_id in summary['active_session_ids'][:10]: # 最多显示10个 + session = self.session_manager.get_session(session_id) + if session: + sid = session_id[:24] if len(session_id) > 24 else session_id + print(f"│ {sid:<24} │ {session['messages_count']:>7} │ {session['total_input_tokens']:>8,} │ {session['total_output_tokens']:>9,} │") + + print("└──────────────────────────┴─────────┴──────────┴───────────┘") + + print() + print(f"📈 Token Statistics") + print(f" Total Input: {summary['total_input_tokens']:,} tokens") + print(f" Total Output: {summary['total_output_tokens']:,} tokens") + if summary['total_reasoning_tokens'] > 0: + print(f" Total Reasoning: {summary['total_reasoning_tokens']:,} tokens") + print(f" Total Cached: {summary['total_cached_tokens']:,} tokens") + print(f" Total Cost: ${summary['total_cost_usd']:.4f}") + + def _display_summary(self): + """显示最终汇总""" + summary = self.session_manager.get_summary() + + print() + print(f"{'=' * 50}") + print(f"📊 Session Monitor - Summary") + print(f"{'=' * 50}") + print() + print(f"📈 Final Statistics") + print(f" Total Sessions: {summary['total_sessions']}") + print(f" Total Input: {summary['total_input_tokens']:,} tokens") + print(f" Total Output: {summary['total_output_tokens']:,} tokens") + if summary['total_reasoning_tokens'] > 0: + print(f" Total Reasoning: {summary['total_reasoning_tokens']:,} tokens") + print(f" Total Cached: {summary['total_cached_tokens']:,} tokens") + print(f" Total Tokens: {summary['total_tokens']:,} tokens") + print(f" Total Cost: ${summary['total_cost_usd']:.4f}") + print(f"{'=' * 50}") + print() + + +# ============================================================================ +# 主程序 +# ============================================================================ + +def main(): + parser = argparse.ArgumentParser( + description="Agent Session Monitor - 实时监控多轮Agent对话的token开销", + formatter_class=argparse.RawDescriptionHelpFormatter, + epilog=""" +示例: + # 监控默认日志 + %(prog)s + + # 监控指定日志文件 + %(prog)s --log-path /var/log/higress/access.log + + # 设置预算为500K tokens + %(prog)s --budget 500000 + + # 监控特定session + %(prog)s --session-key agent:main:discord:channel:1465367993012981988 + """, + allow_abbrev=False + ) + + parser.add_argument( + '--log-path', + default=DEFAULT_LOG_PATH, + help=f'Higress访问日志文件路径（默认: {DEFAULT_LOG_PATH}）' + ) + + parser.add_argument( + '--output-dir', + default=DEFAULT_OUTPUT_DIR, + help=f'Session数据存储目录（默认: {DEFAULT_OUTPUT_DIR}）' + ) + + parser.add_argument( + '--session-key', + help='只监控包含指定session key的日志' + ) + + parser.add_argument( + '--refresh-interval', + type=int, + default=1, + help=f'实时监控刷新间隔（秒，默认: 1）' + ) + + parser.add_argument( + '--state-file', + help='状态文件路径，用于记录已读取的offset（默认: /.state.json）' + ) + + args = parser.parse_args() + + # 初始化组件 + session_manager = SessionManager(output_dir=args.output_dir) + + # 状态文件路径 + state_file = args.state_file or str(Path(args.output_dir) / '.state.json') + + log_parser = LogParser(state_file=state_file) + + print(f"{'=' * 60}") + print(f"🔍 Agent Session Monitor") + print(f"{'=' * 60}") + print() + print(f"📂 Log path: {args.log_path}") + print(f"📁 Output dir: {args.output_dir}") + if args.session_key: + print(f"🔑 Session key filter: {args.session_key}") + print(f"{'=' * 60}") + print() + + # 模式选择：实时监控或单次解析 + if len(sys.argv) == 1: + # 默认模式：实时监控（定时轮询） + print("📺 Mode: Real-time monitoring (polling mode with log rotation support)") + print(f" Refresh interval: {args.refresh_interval} second(s)") + print() + + # 首次解析现有日志文件（包括轮转的文件） + log_parser.parse_rotated_logs(args.log_path, session_manager) + + # 启动实时监控（定时轮询模式） + monitor = RealtimeMonitor( + session_manager, + log_parser=log_parser, + log_path=args.log_path, + refresh_interval=args.refresh_interval + ) + monitor.start() + + else: + # 单次解析模式 + print("📊 Mode: One-time log parsing (with log rotation support)") + print() + log_parser.parse_rotated_logs(args.log_path, session_manager) + + # 显示汇总 + summary = session_manager.get_summary() + print(f"\n{'=' * 50}") + print(f"📊 Session Summary") + print(f"{'=' * 50}") + print() + print(f"📈 Final Statistics") + print(f" Total Sessions: {summary['total_sessions']}") + print(f" Total Input: {summary['total_input_tokens']:,} tokens") + print(f" Total Output: {summary['total_output_tokens']:,} tokens") + if summary['total_reasoning_tokens'] > 0: + print(f" Total Reasoning: {summary['total_reasoning_tokens']:,} tokens") + print(f" Total Cached: {summary['total_cached_tokens']:,} tokens") + print(f" Total Tokens: {summary['total_tokens']:,} tokens") + print(f" Total Cost: ${summary['total_cost_usd']:.4f}") + print(f"{'=' * 50}") + print() + print(f"💾 Session data saved to: {args.output_dir}/") + print(f" Run with --output-dir to specify custom directory") + + +if __name__ == '__main__': + main() diff --git a/.claude/skills/agent-session-monitor/scripts/cli.py b/.claude/skills/agent-session-monitor/scripts/cli.py new file mode 100755 index 000000000..6cc30b09b --- /dev/null +++ b/.claude/skills/agent-session-monitor/scripts/cli.py @@ -0,0 +1,600 @@ +#!/usr/bin/env python3 +""" +Agent Session Monitor CLI - 查询和分析agent对话数据 +支持： +1. 实时查询指定session的完整llm请求和响应 +2. 按模型统计token开销 +3. 按日期统计token开销 +4. 生成FinOps报表 +""" + +import argparse +import json +import sys +from collections import defaultdict +from datetime import datetime, timedelta +from pathlib import Path +from typing import Dict, List, Optional +import re + +# Token定价（单位：美元/1M tokens） +TOKEN_PRICING = { + "Qwen": { + "input": 0.0002, # $0.2/1M + "output": 0.0006, + "cached": 0.0001, # cached tokens通常是input的50% + }, + "Qwen3-rerank": { + "input": 0.0003, + "output": 0.0012, + "cached": 0.00015, + }, + "Qwen-Max": { + "input": 0.0005, + "output": 0.002, + "cached": 0.00025, + }, + "GPT-4": { + "input": 0.003, + "output": 0.006, + "cached": 0.0015, + }, + "GPT-4o": { + "input": 0.0025, + "output": 0.01, + "cached": 0.00125, # GPT-4o prompt caching: 50% discount + }, + "GPT-4-32k": { + "input": 0.01, + "output": 0.03, + "cached": 0.005, + }, + "o1": { + "input": 0.015, + "output": 0.06, + "cached": 0.0075, + "reasoning": 0.06, # o1 reasoning tokens same as output + }, + "o1-mini": { + "input": 0.003, + "output": 0.012, + "cached": 0.0015, + "reasoning": 0.012, + }, + "Claude": { + "input": 0.015, + "output": 0.075, + "cached": 0.0015, # Claude prompt caching: 90% discount + }, + "DeepSeek-R1": { + "input": 0.004, + "output": 0.012, + "reasoning": 0.002, + "cached": 0.002, + } +} + + +class SessionAnalyzer: + """Session数据分析器""" + + def __init__(self, data_dir: str): + self.data_dir = Path(data_dir) + if not self.data_dir.exists(): + raise FileNotFoundError(f"Session data directory not found: {data_dir}") + + def load_session(self, session_id: str) -> Optional[dict]: + """加载指定session的完整数据""" + session_file = self.data_dir / f"{session_id}.json" + if not session_file.exists(): + return None + + with open(session_file, 'r', encoding='utf-8') as f: + return json.load(f) + + def load_all_sessions(self) -> List[dict]: + """加载所有session数据""" + sessions = [] + for session_file in self.data_dir.glob("*.json"): + try: + with open(session_file, 'r', encoding='utf-8') as f: + session = json.load(f) + sessions.append(session) + except Exception as e: + print(f"Warning: Failed to load {session_file}: {e}", file=sys.stderr) + return sessions + + def display_session_detail(self, session_id: str, show_messages: bool = True): + """显示session的详细信息""" + session = self.load_session(session_id) + if not session: + print(f"❌ Session not found: {session_id}") + return + + print(f"\n{'='*70}") + print(f"📊 Session Detail: {session_id}") + print(f"{'='*70}\n") + + # 基本信息 + print(f"🕐 Created: {session['created_at']}") + print(f"🕑 Updated: {session['updated_at']}") + print(f"🤖 Model: {session['model']}") + print(f"💬 Messages: {session['messages_count']}") + print() + + # Token统计 + print(f"📈 Token Statistics:") + + total_input = session['total_input_tokens'] + total_output = session['total_output_tokens'] + total_reasoning = session.get('total_reasoning_tokens', 0) + total_cached = session.get('total_cached_tokens', 0) + + # 区分regular input和cached input + regular_input = total_input - total_cached + + if total_cached > 0: + print(f" Input: {regular_input:>10,} tokens (regular)") + print(f" Cached: {total_cached:>10,} tokens (from cache)") + print(f" Total Input:{total_input:>10,} tokens") + else: + print(f" Input: {total_input:>10,} tokens") + + print(f" Output: {total_output:>10,} tokens") + + if total_reasoning > 0: + print(f" Reasoning: {total_reasoning:>10,} tokens") + + # 总计（不重复计算cached） + total_tokens = total_input + total_output + total_reasoning + print(f" ────────────────────────") + print(f" Total: {total_tokens:>10,} tokens") + print() + + # 成本计算 + cost = self._calculate_cost(session) + print(f"💰 Estimated Cost: ${cost:.8f} USD") + print() + + # 对话轮次 + if show_messages and 'rounds' in session: + print(f"📝 Conversation Rounds ({len(session['rounds'])}):") + print(f"{'─'*70}") + + for i, round_data in enumerate(session['rounds'], 1): + timestamp = round_data.get('timestamp', 'N/A') + input_tokens = round_data.get('input_tokens', 0) + output_tokens = round_data.get('output_tokens', 0) + has_tool_calls = round_data.get('has_tool_calls', False) + response_type = round_data.get('response_type', 'normal') + + print(f"\n Round {i} @ {timestamp}") + print(f" Tokens: {input_tokens:,} in → {output_tokens:,} out") + + if has_tool_calls: + print(f" 🔧 Tool calls: Yes") + + if response_type != 'normal': + print(f" Type: {response_type}") + + # 显示完整的messages（如果有） + if 'messages' in round_data: + messages = round_data['messages'] + print(f" Messages ({len(messages)}):") + for msg in messages[-3:]: # 只显示最后3条 + role = msg.get('role', 'unknown') + content = msg.get('content', '') + content_preview = content[:100] + '...' if len(content) > 100 else content + print(f" [{role}] {content_preview}") + + # 显示question/answer/reasoning（如果有） + if 'question' in round_data: + q = round_data['question'] + q_preview = q[:150] + '...' if len(q) > 150 else q + print(f" ❓ Question: {q_preview}") + + if 'answer' in round_data: + a = round_data['answer'] + a_preview = a[:150] + '...' if len(a) > 150 else a + print(f" ✅ Answer: {a_preview}") + + if 'reasoning' in round_data and round_data['reasoning']: + r = round_data['reasoning'] + r_preview = r[:150] + '...' if len(r) > 150 else r + print(f" 🧠 Reasoning: {r_preview}") + + if 'tool_calls' in round_data and round_data['tool_calls']: + print(f" 🛠️ Tool Calls:") + for tool_call in round_data['tool_calls']: + func_name = tool_call.get('function', {}).get('name', 'unknown') + args = tool_call.get('function', {}).get('arguments', '') + print(f" - {func_name}({args[:80]}...)") + + # 显示token details（如果有） + if round_data.get('input_token_details'): + print(f" 📊 Input Token Details: {round_data['input_token_details']}") + + if round_data.get('output_token_details'): + print(f" 📊 Output Token Details: {round_data['output_token_details']}") + + print(f"\n{'─'*70}") + + print(f"\n{'='*70}\n") + + def _calculate_cost(self, session: dict) -> float: + """计算session的成本""" + model = session.get('model', 'unknown') + pricing = TOKEN_PRICING.get(model, TOKEN_PRICING.get("GPT-4", {})) + + input_tokens = session['total_input_tokens'] + output_tokens = session['total_output_tokens'] + reasoning_tokens = session.get('total_reasoning_tokens', 0) + cached_tokens = session.get('total_cached_tokens', 0) + + # 区分regular input和cached input + regular_input_tokens = input_tokens - cached_tokens + + input_cost = regular_input_tokens * pricing.get('input', 0) / 1000000 + output_cost = output_tokens * pricing.get('output', 0) / 1000000 + + reasoning_cost = 0 + if 'reasoning' in pricing and reasoning_tokens > 0: + reasoning_cost = reasoning_tokens * pricing['reasoning'] / 1000000 + + cached_cost = 0 + if 'cached' in pricing and cached_tokens > 0: + cached_cost = cached_tokens * pricing['cached'] / 1000000 + + return input_cost + output_cost + reasoning_cost + cached_cost + + def stats_by_model(self) -> Dict[str, dict]: + """按模型统计token开销""" + sessions = self.load_all_sessions() + + stats = defaultdict(lambda: { + 'session_count': 0, + 'total_input': 0, + 'total_output': 0, + 'total_reasoning': 0, + 'total_cost': 0.0 + }) + + for session in sessions: + model = session.get('model', 'unknown') + stats[model]['session_count'] += 1 + stats[model]['total_input'] += session['total_input_tokens'] + stats[model]['total_output'] += session['total_output_tokens'] + stats[model]['total_reasoning'] += session.get('total_reasoning_tokens', 0) + stats[model]['total_cost'] += self._calculate_cost(session) + + return dict(stats) + + def stats_by_date(self, days: int = 30) -> Dict[str, dict]: + """按日期统计token开销（最近N天）""" + sessions = self.load_all_sessions() + + stats = defaultdict(lambda: { + 'session_count': 0, + 'total_input': 0, + 'total_output': 0, + 'total_reasoning': 0, + 'total_cost': 0.0, + 'models': set() + }) + + cutoff_date = datetime.now() - timedelta(days=days) + + for session in sessions: + created_at = datetime.fromisoformat(session['created_at']) + if created_at < cutoff_date: + continue + + date_key = created_at.strftime('%Y-%m-%d') + stats[date_key]['session_count'] += 1 + stats[date_key]['total_input'] += session['total_input_tokens'] + stats[date_key]['total_output'] += session['total_output_tokens'] + stats[date_key]['total_reasoning'] += session.get('total_reasoning_tokens', 0) + stats[date_key]['total_cost'] += self._calculate_cost(session) + stats[date_key]['models'].add(session.get('model', 'unknown')) + + # 转换sets为lists以便JSON序列化 + for date_key in stats: + stats[date_key]['models'] = list(stats[date_key]['models']) + + return dict(stats) + + def display_model_stats(self): + """显示按模型的统计""" + stats = self.stats_by_model() + + print(f"\n{'='*80}") + print(f"📊 Statistics by Model") + print(f"{'='*80}\n") + + print(f"{'Model':<20} {'Sessions':<10} {'Input':<15} {'Output':<15} {'Cost (USD)':<12}") + print(f"{'─'*80}") + + # 按成本降序排列 + sorted_models = sorted(stats.items(), key=lambda x: x[1]['total_cost'], reverse=True) + + for model, data in sorted_models: + print(f"{model:<20} " + f"{data['session_count']:<10} " + f"{data['total_input']:>12,} " + f"{data['total_output']:>12,} " + f"${data['total_cost']:>10.6f}") + + # 总计 + total_sessions = sum(d['session_count'] for d in stats.values()) + total_input = sum(d['total_input'] for d in stats.values()) + total_output = sum(d['total_output'] for d in stats.values()) + total_cost = sum(d['total_cost'] for d in stats.values()) + + print(f"{'─'*80}") + print(f"{'TOTAL':<20} " + f"{total_sessions:<10} " + f"{total_input:>12,} " + f"{total_output:>12,} " + f"${total_cost:>10.6f}") + + print(f"\n{'='*80}\n") + + def display_date_stats(self, days: int = 30): + """显示按日期的统计""" + stats = self.stats_by_date(days) + + print(f"\n{'='*80}") + print(f"📊 Statistics by Date (Last {days} days)") + print(f"{'='*80}\n") + + print(f"{'Date':<12} {'Sessions':<10} {'Input':<15} {'Output':<15} {'Cost (USD)':<12} {'Models':<20}") + print(f"{'─'*80}") + + # 按日期升序排列 + sorted_dates = sorted(stats.items()) + + for date, data in sorted_dates: + models_str = ', '.join(data['models'][:3]) # 最多显示3个模型 + if len(data['models']) > 3: + models_str += f" +{len(data['models'])-3}" + + print(f"{date:<12} " + f"{data['session_count']:<10} " + f"{data['total_input']:>12,} " + f"{data['total_output']:>12,} " + f"${data['total_cost']:>10.4f} " + f"{models_str}") + + # 总计 + total_sessions = sum(d['session_count'] for d in stats.values()) + total_input = sum(d['total_input'] for d in stats.values()) + total_output = sum(d['total_output'] for d in stats.values()) + total_cost = sum(d['total_cost'] for d in stats.values()) + + print(f"{'─'*80}") + print(f"{'TOTAL':<12} " + f"{total_sessions:<10} " + f"{total_input:>12,} " + f"{total_output:>12,} " + f"${total_cost:>10.4f}") + + print(f"\n{'='*80}\n") + + def list_sessions(self, limit: int = 20, sort_by: str = 'updated'): + """列出所有session""" + sessions = self.load_all_sessions() + + # 排序 + if sort_by == 'updated': + sessions.sort(key=lambda s: s.get('updated_at', ''), reverse=True) + elif sort_by == 'cost': + sessions.sort(key=lambda s: self._calculate_cost(s), reverse=True) + elif sort_by == 'tokens': + sessions.sort(key=lambda s: s['total_input_tokens'] + s['total_output_tokens'], reverse=True) + + print(f"\n{'='*100}") + print(f"📋 Sessions (sorted by {sort_by}, showing {min(limit, len(sessions))} of {len(sessions)})") + print(f"{'='*100}\n") + + print(f"{'Session ID':<30} {'Updated':<20} {'Model':<15} {'Msgs':<6} {'Tokens':<12} {'Cost':<10}") + print(f"{'─'*100}") + + for session in sessions[:limit]: + session_id = session['session_id'][:28] + '..' if len(session['session_id']) > 30 else session['session_id'] + updated = session.get('updated_at', 'N/A')[:19] + model = session.get('model', 'unknown')[:13] + msg_count = session.get('messages_count', 0) + total_tokens = session['total_input_tokens'] + session['total_output_tokens'] + cost = self._calculate_cost(session) + + print(f"{session_id:<30} {updated:<20} {model:<15} {msg_count:<6} {total_tokens:>10,} ${cost:>8.4f}") + + print(f"\n{'='*100}\n") + + def export_finops_report(self, output_file: str, format: str = 'json'): + """导出FinOps报表""" + model_stats = self.stats_by_model() + date_stats = self.stats_by_date(30) + + report = { + 'generated_at': datetime.now().isoformat(), + 'summary': { + 'total_sessions': sum(d['session_count'] for d in model_stats.values()), + 'total_input_tokens': sum(d['total_input'] for d in model_stats.values()), + 'total_output_tokens': sum(d['total_output'] for d in model_stats.values()), + 'total_cost_usd': sum(d['total_cost'] for d in model_stats.values()), + }, + 'by_model': model_stats, + 'by_date': date_stats, + } + + output_path = Path(output_file) + + if format == 'json': + with open(output_path, 'w', encoding='utf-8') as f: + json.dump(report, f, ensure_ascii=False, indent=2) + print(f"✅ FinOps report exported to: {output_path}") + + elif format == 'csv': + import csv + + # 按模型导出CSV + model_csv = output_path.with_suffix('.model.csv') + with open(model_csv, 'w', newline='', encoding='utf-8') as f: + writer = csv.writer(f) + writer.writerow(['Model', 'Sessions', 'Input Tokens', 'Output Tokens', 'Cost (USD)']) + for model, data in model_stats.items(): + writer.writerow([ + model, + data['session_count'], + data['total_input'], + data['total_output'], + f"{data['total_cost']:.6f}" + ]) + + # 按日期导出CSV + date_csv = output_path.with_suffix('.date.csv') + with open(date_csv, 'w', newline='', encoding='utf-8') as f: + writer = csv.writer(f) + writer.writerow(['Date', 'Sessions', 'Input Tokens', 'Output Tokens', 'Cost (USD)', 'Models']) + for date, data in sorted(date_stats.items()): + writer.writerow([ + date, + data['session_count'], + data['total_input'], + data['total_output'], + f"{data['total_cost']:.6f}", + ', '.join(data['models']) + ]) + + print(f"✅ FinOps report exported to:") + print(f" Model stats: {model_csv}") + print(f" Date stats: {date_csv}") + + +def main(): + parser = argparse.ArgumentParser( + description="Agent Session Monitor CLI - 查询和分析agent对话数据", + formatter_class=argparse.RawDescriptionHelpFormatter, + epilog=""" +Commands: + show 显示session的详细信息 + list 列出所有session + stats-model 按模型统计token开销 + stats-date 按日期统计token开销（默认30天） + export 导出FinOps报表 + +Examples: + # 查看特定session的详细对话 + %(prog)s show agent:main:discord:channel:1465367993012981988 + + # 列出最近20个session（按更新时间） + %(prog)s list + + # 列出token开销最高的10个session + %(prog)s list --sort-by cost --limit 10 + + # 按模型统计token开销 + %(prog)s stats-model + + # 按日期统计token开销（最近7天） + %(prog)s stats-date --days 7 + + # 导出FinOps报表（JSON格式） + %(prog)s export finops-report.json + + # 导出FinOps报表（CSV格式） + %(prog)s export finops-report --format csv + """ + ) + + parser.add_argument( + 'command', + choices=['show', 'list', 'stats-model', 'stats-date', 'export'], + help='命令' + ) + + parser.add_argument( + 'args', + nargs='*', + help='命令参数（例如：session-id或输出文件名）' + ) + + parser.add_argument( + '--data-dir', + default='./sessions', + help='Session数据目录（默认: ./sessions）' + ) + + parser.add_argument( + '--limit', + type=int, + default=20, + help='list命令的结果限制（默认: 20）' + ) + + parser.add_argument( + '--sort-by', + choices=['updated', 'cost', 'tokens'], + default='updated', + help='list命令的排序方式（默认: updated）' + ) + + parser.add_argument( + '--days', + type=int, + default=30, + help='stats-date命令的天数（默认: 30）' + ) + + parser.add_argument( + '--format', + choices=['json', 'csv'], + default='json', + help='export命令的输出格式（默认: json）' + ) + + parser.add_argument( + '--no-messages', + action='store_true', + help='show命令：不显示对话内容' + ) + + args = parser.parse_args() + + try: + analyzer = SessionAnalyzer(args.data_dir) + + if args.command == 'show': + if not args.args: + parser.error("show命令需要session-id参数") + session_id = args.args[0] + analyzer.display_session_detail(session_id, show_messages=not args.no_messages) + + elif args.command == 'list': + analyzer.list_sessions(limit=args.limit, sort_by=args.sort_by) + + elif args.command == 'stats-model': + analyzer.display_model_stats() + + elif args.command == 'stats-date': + analyzer.display_date_stats(days=args.days) + + elif args.command == 'export': + if not args.args: + parser.error("export命令需要输出文件名参数") + output_file = args.args[0] + analyzer.export_finops_report(output_file, format=args.format) + + except FileNotFoundError as e: + print(f"❌ Error: {e}", file=sys.stderr) + sys.exit(1) + except Exception as e: + print(f"❌ Unexpected error: {e}", file=sys.stderr) + import traceback + traceback.print_exc() + sys.exit(1) + + +if __name__ == '__main__': + main() diff --git a/.claude/skills/agent-session-monitor/scripts/webserver.py b/.claude/skills/agent-session-monitor/scripts/webserver.py new file mode 100755 index 000000000..7d6a90967 --- /dev/null +++ b/.claude/skills/agent-session-monitor/scripts/webserver.py @@ -0,0 +1,755 @@ +#!/usr/bin/env python3 +""" +Agent Session Monitor - Web Server +提供浏览器访问的观测界面 +""" + +import argparse +import json +import sys +from pathlib import Path +from http.server import HTTPServer, BaseHTTPRequestHandler +from urllib.parse import urlparse, parse_qs +from collections import defaultdict +from datetime import datetime, timedelta +import re + +# 添加父目录到path以导入cli模块 +sys.path.insert(0, str(Path(__file__).parent.parent)) + +try: + from scripts.cli import SessionAnalyzer, TOKEN_PRICING +except ImportError: + # 如果导入失败，定义简单版本 + TOKEN_PRICING = { + "Qwen3-rerank": {"input": 0.0003, "output": 0.0012}, + "DeepSeek-R1": {"input": 0.004, "output": 0.012, "reasoning": 0.002}, + } + + +class SessionMonitorHandler(BaseHTTPRequestHandler): + """HTTP请求处理器""" + + def __init__(self, *args, data_dir=None, **kwargs): + self.data_dir = Path(data_dir) if data_dir else Path("./sessions") + super().__init__(*args, **kwargs) + + def do_GET(self): + """处理GET请求""" + parsed_path = urlparse(self.path) + path = parsed_path.path + query = parse_qs(parsed_path.query) + + if path == '/' or path == '/index.html': + self.serve_index() + elif path == '/session': + session_id = query.get('id', [None])[0] + if session_id: + self.serve_session_detail(session_id) + else: + self.send_error(400, "Missing session id") + elif path == '/api/sessions': + self.serve_api_sessions() + elif path == '/api/session': + session_id = query.get('id', [None])[0] + if session_id: + self.serve_api_session(session_id) + else: + self.send_error(400, "Missing session id") + elif path == '/api/stats': + self.serve_api_stats() + else: + self.send_error(404, "Not Found") + + def serve_index(self): + """首页 - 总览""" + html = self.generate_index_html() + self.send_html(html) + + def serve_session_detail(self, session_id: str): + """Session详情页""" + html = self.generate_session_html(session_id) + self.send_html(html) + + def serve_api_sessions(self): + """API: 获取所有session列表""" + sessions = self.load_all_sessions() + + # 简化数据 + data = [] + for session in sessions: + data.append({ + 'session_id': session['session_id'], + 'model': session.get('model', 'unknown'), + 'messages_count': session.get('messages_count', 0), + 'total_tokens': session['total_input_tokens'] + session['total_output_tokens'], + 'updated_at': session.get('updated_at', ''), + 'cost': self.calculate_cost(session) + }) + + # 按更新时间降序排序 + data.sort(key=lambda x: x['updated_at'], reverse=True) + + self.send_json(data) + + def serve_api_session(self, session_id: str): + """API: 获取指定session的详细数据""" + session = self.load_session(session_id) + if session: + session['cost'] = self.calculate_cost(session) + self.send_json(session) + else: + self.send_error(404, "Session not found") + + def serve_api_stats(self): + """API: 获取统计数据""" + sessions = self.load_all_sessions() + + # 按模型统计 + by_model = defaultdict(lambda: { + 'count': 0, + 'input_tokens': 0, + 'output_tokens': 0, + 'cost': 0.0 + }) + + # 按日期统计 + by_date = defaultdict(lambda: { + 'count': 0, + 'input_tokens': 0, + 'output_tokens': 0, + 'cost': 0.0, + 'models': set() + }) + + total_cost = 0.0 + + for session in sessions: + model = session.get('model', 'unknown') + cost = self.calculate_cost(session) + total_cost += cost + + # 按模型 + by_model[model]['count'] += 1 + by_model[model]['input_tokens'] += session['total_input_tokens'] + by_model[model]['output_tokens'] += session['total_output_tokens'] + by_model[model]['cost'] += cost + + # 按日期 + created_at = session.get('created_at', '') + date_key = created_at[:10] if len(created_at) >= 10 else 'unknown' + by_date[date_key]['count'] += 1 + by_date[date_key]['input_tokens'] += session['total_input_tokens'] + by_date[date_key]['output_tokens'] += session['total_output_tokens'] + by_date[date_key]['cost'] += cost + by_date[date_key]['models'].add(model) + + # 转换sets为lists + for date in by_date: + by_date[date]['models'] = list(by_date[date]['models']) + + stats = { + 'total_sessions': len(sessions), + 'total_cost': total_cost, + 'by_model': dict(by_model), + 'by_date': dict(sorted(by_date.items(), reverse=True)) + } + + self.send_json(stats) + + def load_session(self, session_id: str): + """加载指定session""" + session_file = self.data_dir / f"{session_id}.json" + if session_file.exists(): + with open(session_file, 'r', encoding='utf-8') as f: + return json.load(f) + return None + + def load_all_sessions(self): + """加载所有session""" + sessions = [] + for session_file in self.data_dir.glob("*.json"): + try: + with open(session_file, 'r', encoding='utf-8') as f: + sessions.append(json.load(f)) + except Exception as e: + print(f"Warning: Failed to load {session_file}: {e}", file=sys.stderr) + return sessions + + def calculate_cost(self, session: dict) -> float: + """计算session成本""" + model = session.get('model', 'unknown') + pricing = TOKEN_PRICING.get(model, TOKEN_PRICING.get("GPT-4", {"input": 0.003, "output": 0.006})) + + input_tokens = session['total_input_tokens'] + output_tokens = session['total_output_tokens'] + reasoning_tokens = session.get('total_reasoning_tokens', 0) + cached_tokens = session.get('total_cached_tokens', 0) + + # 区分regular input和cached input + regular_input_tokens = input_tokens - cached_tokens + + input_cost = regular_input_tokens * pricing.get('input', 0) / 1000000 + output_cost = output_tokens * pricing.get('output', 0) / 1000000 + + reasoning_cost = 0 + if 'reasoning' in pricing and reasoning_tokens > 0: + reasoning_cost = reasoning_tokens * pricing['reasoning'] / 1000000 + + cached_cost = 0 + if 'cached' in pricing and cached_tokens > 0: + cached_cost = cached_tokens * pricing['cached'] / 1000000 + + return input_cost + output_cost + reasoning_cost + cached_cost + + def send_html(self, html: str): + """发送HTML响应""" + self.send_response(200) + self.send_header('Content-type', 'text/html; charset=utf-8') + self.end_headers() + self.wfile.write(html.encode('utf-8')) + + def send_json(self, data): + """发送JSON响应""" + self.send_response(200) + self.send_header('Content-type', 'application/json; charset=utf-8') + self.send_header('Access-Control-Allow-Origin', '*') + self.end_headers() + self.wfile.write(json.dumps(data, ensure_ascii=False, indent=2).encode('utf-8')) + + def generate_index_html(self) -> str: + """生成首页HTML""" + return ''' + + + + + Agent Session Monitor + + + +
+
+

🔍 Agent Session Monitor

+

实时观测Clawdbot对话过程和Token开销

+
+ +
+
+
总会话数
+
-
+
+
+
总Token消耗
+
-
+
+
+
总成本
+
-
+
+
+ +
+

📊 最近会话

+ +
+
加载中...
+
+
+ +
+

📈 按模型统计

+
+
加载中...
+
+
+
+ + + +''' + + def generate_session_html(self, session_id: str) -> str: + """生成Session详情页HTML""" + session = self.load_session(session_id) + if not session: + return f'

Session not found: {session_id}

' + + cost = self.calculate_cost(session) + + # 生成对话轮次HTML + rounds_html = [] + for r in session.get('rounds', []): + messages_html = '' + if r.get('messages'): + messages_html = '
' + for msg in r['messages'][-5:]: # 最多显示5条 + role = msg.get('role', 'unknown') + content = msg.get('content', '') + messages_html += f'
[{role}] {self.escape_html(content)}
' + messages_html += '
' + + tool_calls_html = '' + if r.get('tool_calls'): + tool_calls_html = '
🛠️ Tool Calls:
    ' + for tc in r['tool_calls']: + func_name = tc.get('function', {}).get('name', 'unknown') + tool_calls_html += f'
  • {func_name}()
    • ' + tool_calls_html += '
' + + # Token详情显示 + token_details_html = '' + if r.get('input_token_details') or r.get('output_token_details'): + token_details_html = '
📊 Token Details:
    ' + if r.get('input_token_details'): + token_details_html += f'
  • Input: {r["input_token_details"]}
    • ' + if r.get('output_token_details'): + token_details_html += f'
  • Output: {r["output_token_details"]}
    • ' + token_details_html += '
' + + # Token类型标签 + token_badges = '' + if r.get('cached_tokens', 0) > 0: + token_badges += f' 📦 {r["cached_tokens"]:,} cached' + if r.get('reasoning_tokens', 0) > 0: + token_badges += f' 🧠 {r["reasoning_tokens"]:,} reasoning' + + rounds_html.append(f''' +
+
+ Round {r['round']} + {r['timestamp']} + {r['input_tokens']:,} in → {r['output_tokens']:,} out{token_badges} +
+ {messages_html} + {f'
❓ Question: {self.escape_html(r.get("question", ""))}
' if r.get('question') else ''} + {f'
✅ Answer: {self.escape_html(r.get("answer", ""))}
' if r.get('answer') else ''} + {f'
🧠 Reasoning: {self.escape_html(r.get("reasoning", ""))}
' if r.get('reasoning') else ''} + {tool_calls_html} + {token_details_html} +
+ ''') + + return f''' + + + + + {session_id} - Session Monitor + + + +
+
+ ← 返回列表 +

📊 Session Detail

+

{session_id}

+ +
+
+
模型
+
{session.get('model', 'unknown')}
+
+
+
消息数
+
{session.get('messages_count', 0)}
+
+
+
总Token
+
{session['total_input_tokens'] + session['total_output_tokens']:,}
+
+
+
成本
+
${cost:.6f}
+
+
+
+ +
+

💬 对话记录 ({len(session.get('rounds', []))} 轮)

+ {"".join(rounds_html) if rounds_html else '

暂无对话记录

'} +
+
+ +''' + + def escape_html(self, text: str) -> str: + """转义HTML特殊字符""" + return (text.replace('&', '&') + .replace('<', '<') + .replace('>', '>') + .replace('"', '"') + .replace("'", ''')) + + def log_message(self, format, *args): + """重写日志方法，简化输出""" + pass # 不打印每个请求 + + +def create_handler(data_dir): + """创建带数据目录的处理器""" + def handler(*args, **kwargs): + return SessionMonitorHandler(*args, data_dir=data_dir, **kwargs) + return handler + + +def main(): + parser = argparse.ArgumentParser( + description="Agent Session Monitor - Web Server", + formatter_class=argparse.RawDescriptionHelpFormatter + ) + + parser.add_argument( + '--data-dir', + default='./sessions', + help='Session数据目录（默认: ./sessions）' + ) + + parser.add_argument( + '--port', + type=int, + default=8888, + help='HTTP服务器端口（默认: 8888）' + ) + + parser.add_argument( + '--host', + default='0.0.0.0', + help='HTTP服务器地址（默认: 0.0.0.0）' + ) + + args = parser.parse_args() + + # 检查数据目录是否存在 + data_dir = Path(args.data_dir) + if not data_dir.exists(): + print(f"❌ Error: Data directory not found: {data_dir}") + print(f" Please run main.py first to generate session data.") + sys.exit(1) + + # 创建HTTP服务器 + handler_class = create_handler(args.data_dir) + server = HTTPServer((args.host, args.port), handler_class) + + print(f"{'=' * 60}") + print(f"🌐 Agent Session Monitor - Web Server") + print(f"{'=' * 60}") + print() + print(f"📂 Data directory: {args.data_dir}") + print(f"🌍 Server address: http://{args.host}:{args.port}") + print() + print(f"✅ Server started. Press Ctrl+C to stop.") + print(f"{'=' * 60}") + print() + + try: + server.serve_forever() + except KeyboardInterrupt: + print("\n\n👋 Shutting down server...") + server.shutdown() + + +if __name__ == '__main__': + main()