高级功能

掌握 OpenClaw 的高级特性,释放 AI Agent 的全部潜力。

文件管理

Agent 可以读写文件系统:

file_system:
  workspace: "/workspace"

  permissions:
    read:
      - "/workspace/**"
      - "/data/**"

    write:
      - "/workspace/**"
      - "/output/**"

    execute:
      - "/workspace/scripts/**"

  max_file_size: "100MB"

  allowed_extensions:
    - .txt
    - .md
    - .json
    - .py
    - .js

代码执行沙箱

安全地执行用户代码:

sandbox:
  enabled: true

  python:
    enabled: true
    version: "3.11"
    timeout: 60s
    max_memory: "1GB"

    allowed_packages:
      - numpy
      - pandas
      - matplotlib
      - requests

  javascript:
    enabled: true
    runtime: "node"
    version: "20"
    timeout: 30s

使用示例:

# 用户: 帮我用 Python 分析这个 CSV 文件

import pandas as pd
df = pd.read_csv('/workspace/data.csv')
print(df.describe())

RAG 知识库

接入向量数据库,实现 RAG:

knowledge_base:
  enabled: true

  vector_db:
    type: "qdrant"
    host: "localhost"
    port: 6333

  embedding:
    model: "text-embedding-3-small"
    dimension: 1536

  retrieval:
    top_k: 5
    min_score: 0.7

  sources:
    - name: "公司文档"
      path: "/knowledge/company_docs"
      sync_interval: "1h"

Agent 会自动从知识库检索相关信息回答问题。

API 集成

调用外部 API 扩展能力:

integrations:
  - name: "Weather API"
    type: "rest"
    base_url: "https://api.weather.com"
    auth:
      type: "api_key"
      header: "X-API-Key"
      value: "${WEATHER_API_KEY}"

    endpoints:
      - name: "get_weather"
        method: "GET"
        path: "/current"
        params:
          city: "{city}"

  - name: "GitHub API"
    type: "graphql"
    url: "https://api.github.com/graphql"
    auth:
      type: "bearer"
      token: "${GITHUB_TOKEN}"

Agent 可以调用这些 API 获取实时数据。