OpenClaw对话记录导出与数据分析

前言

对话数据是 AI Agent 运营中最宝贵的资产之一。通过分析对话记录，你可以了解用户的真实需求、发现 Agent 的不足之处、优化提示词效果、计算成本消耗。OpenClaw 以 JSONL 格式存储所有会话数据，提供了灵活的导出工具和丰富的分析维度。

本文将详细介绍 JSONL 会话文件的结构、导出方法以及实际的数据分析场景。

JSONL 会话文件结构

文件位置

OpenClaw 的会话数据默认存储在以下路径：

~/.openclaw/agents/<agentId>/sessions/
├── session-abc123.jsonl
├── session-def456.jsonl
└── session-ghi789.jsonl

消息格式

每行是一个独立的 JSON 对象，代表一条消息：

{"id":"msg_001","parentId":null,"role":"user","content":"你好","timestamp":1710400000,"metadata":{"userId":"user_123","platform":"telegram","channelType":"dm"}}
{"id":"msg_002","parentId":"msg_001","role":"assistant","content":"你好！有什么可以帮你的？","timestamp":1710400002,"metadata":{"model":"claude-sonnet-4-20250514","inputTokens":45,"outputTokens":12}}
{"id":"msg_003","parentId":"msg_002","role":"user","content":"帮我写一段Python代码","timestamp":1710400010,"metadata":{"userId":"user_123"}}
{"id":"msg_004","parentId":"msg_003","role":"assistant","content":"好的，这是一段示例代码...","timestamp":1710400015,"metadata":{"model":"claude-sonnet-4-20250514","inputTokens":120,"outputTokens":280,"toolsUsed":["run_code"]}}

消息字段说明

元数据字段

{
  "metadata": {
    "userId": "user_123",
    "platform": "telegram",
    "channelType": "dm",
    "channelId": "chat_456",
    "model": "claude-sonnet-4-20250514",
    "inputTokens": 150,
    "outputTokens": 320,
    "totalTokens": 470,
    "latencyMs": 2340,
    "toolsUsed": ["search", "run_code"],
    "costUsd": 0.0023
  }
}

导出对话记录

命令行导出

# 导出某Agent的所有会话
openclaw export --agent my-agent --output ./export/

# 导出指定时间范围
openclaw export --agent my-agent \
  --from "2026-03-01" --to "2026-03-14" \
  --output ./export/march.jsonl

# 导出为CSV格式
openclaw export --agent my-agent \
  --format csv --output ./export/conversations.csv

# 导出为JSON格式（带完整的树形结构）
openclaw export --agent my-agent \
  --format json --output ./export/conversations.json

# 仅导出特定用户的对话
openclaw export --agent my-agent \
  --user-id "user_123" --output ./export/user123.jsonl

# 仅导出特定平台的对话
openclaw export --agent my-agent \
  --platform telegram --output ./export/telegram.jsonl

API 导出

# 通过API导出
curl -X GET "http://localhost:3000/api/v1/export/conversations" \
  -H "Authorization: Bearer sk-openclaw-xxx" \
  -G -d "agentId=my-agent" \
  -d "from=2026-03-01" \
  -d "to=2026-03-14" \
  -d "format=jsonl" \
  -o conversations.jsonl

Dashboard 导出

OpenClaw Web Dashboard 提供了可视化的导出界面：

1. 进入 Dashboard → 会话管理
2. 设置筛选条件（时间范围、Agent、平台等）
3. 点击"导出"按钮
4. 选择格式（JSONL / CSV / JSON）
5. 下载文件

数据分析实践

Python 分析脚本

基础数据加载

import json
from datetime import datetime
from collections import Counter, defaultdict

def load_sessions(filepath):
    messages = []
    with open(filepath, "r", encoding="utf-8") as f:
        for line in f:
            if line.strip():
                messages.append(json.loads(line))
    return messages

messages = load_sessions("./export/conversations.jsonl")
print(f"总消息数: {len(messages)}")

分析一：Token 消耗统计

def analyze_token_usage(messages):
    total_input = 0
    total_output = 0
    daily_usage = defaultdict(lambda: {"input": 0, "output": 0})

    for msg in messages:
        if msg["role"] == "assistant" and "metadata" in msg:
            meta = msg["metadata"]
            input_t = meta.get("inputTokens", 0)
            output_t = meta.get("outputTokens", 0)
            total_input += input_t
            total_output += output_t

            date = datetime.fromtimestamp(msg["timestamp"]).strftime("%Y-%m-%d")
            daily_usage[date]["input"] += input_t
            daily_usage[date]["output"] += output_t

    print(f"总输入Token: {total_input:,}")
    print(f"总输出Token: {total_output:,}")
    print(f"日均输入Token: {total_input // max(len(daily_usage), 1):,}")

    return daily_usage

usage = analyze_token_usage(messages)

分析二：用户活跃度

def analyze_user_activity(messages):
    user_messages = Counter()
    user_sessions = defaultdict(set)

    for msg in messages:
        if msg["role"] == "user" and "metadata" in msg:
            uid = msg["metadata"].get("userId", "unknown")
            user_messages[uid] += 1

    print("Top 10 活跃用户:")
    for uid, count in user_messages.most_common(10):
        print(f"  {uid}: {count} 条消息")

    return user_messages

activity = analyze_user_activity(messages)

分析三：常见问题分类

def categorize_topics(messages):
    """简单的关键词分类"""
    categories = {
        "技术问题": ["代码", "bug", "错误", "部署", "配置"],
        "产品咨询": ["价格", "功能", "对比", "试用"],
        "使用帮助": ["怎么", "如何", "教程", "步骤"],
        "反馈建议": ["建议", "希望", "改进", "不好用"]
    }

    results = Counter()
    for msg in messages:
        if msg["role"] == "user":
            content = msg["content"]
            for category, keywords in categories.items():
                if any(kw in content for kw in keywords):
                    results[category] += 1
                    break

    return results

topics = categorize_topics(messages)
for topic, count in topics.most_common():
    print(f"  {topic}: {count}")

分析四：响应延迟分布

def analyze_latency(messages):
    latencies = []
    for msg in messages:
        if msg["role"] == "assistant" and "metadata" in msg:
            latency = msg["metadata"].get("latencyMs")
            if latency:
                latencies.append(latency)

    if latencies:
        latencies.sort()
        print(f"平均延迟: {sum(latencies)/len(latencies):.0f}ms")
        print(f"P50延迟: {latencies[len(latencies)//2]:.0f}ms")
        print(f"P95延迟: {latencies[int(len(latencies)*0.95)]:.0f}ms")
        print(f"P99延迟: {latencies[int(len(latencies)*0.99)]:.0f}ms")

analyze_latency(messages)

分析五：工具使用统计

def analyze_tool_usage(messages):
    tool_counts = Counter()
    for msg in messages:
        if msg["role"] == "assistant" and "metadata" in msg:
            tools = msg["metadata"].get("toolsUsed", [])
            for tool in tools:
                tool_counts[tool] += 1

    print("工具使用频率:")
    for tool, count in tool_counts.most_common():
        print(f"  {tool}: {count} 次")

analyze_tool_usage(messages)

数据可视化

使用 matplotlib 绘图

import matplotlib.pyplot as plt

def plot_daily_usage(daily_usage):
    dates = sorted(daily_usage.keys())
    input_tokens = [daily_usage[d]["input"] for d in dates]
    output_tokens = [daily_usage[d]["output"] for d in dates]

    fig, ax = plt.subplots(figsize=(12, 6))
    ax.bar(dates, input_tokens, label="输入Token", alpha=0.7)
    ax.bar(dates, output_tokens, bottom=input_tokens,
           label="输出Token", alpha=0.7)
    ax.set_xlabel("日期")
    ax.set_ylabel("Token数")
    ax.set_title("每日Token消耗趋势")
    ax.legend()
    plt.xticks(rotation=45)
    plt.tight_layout()
    plt.savefig("daily_token_usage.png", dpi=150)

plot_daily_usage(usage)

隐私与合规

数据脱敏导出

# 导出时自动脱敏
openclaw export --agent my-agent \
  --anonymize \
  --redact-patterns "phone,email,id_card" \
  --output ./export/anonymized.jsonl

数据保留策略

{
  storage: {
    retention: {
      // 会话数据保留天数
      sessionTTL: 90,  // 90天后自动清理
      // 导出数据保留天数
      exportTTL: 30,
      // 清理前是否自动归档
      archiveBeforeDelete: true,
      archiveDir: "./archive/"
    }
  }
}

小结

OpenClaw 的 JSONL 会话存储格式既简单又强大——每行一条消息，JSON 格式易于解析，树形结构完整记录对话分支。通过命令行、API 或 Dashboard 导出数据后，你可以使用 Python 等工具进行深入分析，从 Token 消耗、用户活跃度、话题分布到响应延迟，全方位了解 AI Agent 的运行状况。这些数据洞察将帮助你持续优化 Agent 的表现和成本效率。