2024年10月7日
智能体实战之autogen：智能体生成shell脚本并执行
内容目录
使用智能体生成脚本并执行
智能体如何使用工具，是研究智能体非常重要的一部分。通过使用脚本，可以赋予智能体使用Linux系统的各式各样的命令行工具来执行特定任务，本文提供非常简单的例子，使用autogen生成脚本并执行用户指定的任务。
智能体介绍
参考文章：什么是智能体
Autogen
AutoGen是一个开源编程框架，用于构建AI智能体并促进多个智能体之间的合作以解决任务。AutoGen旨在提供一个易于使用且灵活的框架，以加速智能体AI的开发和研究，类似于深度学习中的PyTorch。它提供的功能包括能够与其他智能体对话的智能体、支持LLM和工具使用、自治和人机协作工作流程以及多智能体对话模式。
主要功能
- AutoGen使得基于多智能体对话构建下一代LLM应用程序变得轻松便捷。它简化了复杂LLM工作流程的协调、自动化和优化，最大化LLM模型的性能，并克服其弱点。
- 它支持多种对话模式以应对复杂工作流程。通过可定制且可对话的智能体，开发人员可以使用AutoGen构建与对话自主性、智能体数量和智能体对话拓扑相关的各种对话模式。
- 它提供了一系列具有不同复杂性的工作系统。这些系统涵盖了来自各种领域和复杂性的广泛应用，展示了AutoGen如何轻松支持多样的对话模式。
环境准备
- Linux or Windows WSL
- python 3.10+
- 安装 anaconda
本文使用anaconda构建python虚拟环境：
#创建autogen的虚拟环境
conda create --name autogen python=3.11
#进入该虚拟环境
conda activate autogen
# 安装autogen
pip install autogen-agentchat~=0.2
导入相关的依赖库
import os
import tempfile
from autogen import ConversableAgent
from autogen.coding import LocalCommandLineCodeExecutor
构建系统提示词
使用系统提示词定义智能体的角色，指导行为，提供一定的背景信息，同时也可以约束智能体，当用户行为超出智能体能力边界时，提示用户。
以下提示词指导智能体通过生成脚本，来收集Linux系统的一些信息。该提示词为one-shot提示词。提供代码块的样例给大模型。
以下系统提示词将传递给代码生成智能体，指导代码生成智能体生成脚本代码。
script_executor_message = """You are shell script assistant, a expert of linux and shell script, and you will generate shell script and get information according to user's prompt.
Gnerate a script according user's requirments with below format:
#```bash 此处因为显示格式的原因加了注释，实际使用时请删除#
#!/bin/bash
# query all files information in current directory
ls -la
#``` 此处因为显示格式的原因加了注释，实际使用时请删除#
Always thinking step by step to about users questions, make sure your answer is correct and helpful.
If user did not ask question about Linux system, then please response like:
Please ask me question about Linux system, I only could help you on that.
"""
创建和脚本执行器
代码执行器是autogen智能体的一个组件，在创建autogen智能体实例之前需要先初始化代码执行器，并作为参数传递给autogen智能体的实例，代码执行器主要用来执行代码。
# Create a temporary directory to store the code files.
temp_dir = tempfile.TemporaryDirectory()
# Create a local command line code executor.
script_executor = LocalCommandLineCodeExecutor(
timeout=10, # Timeout for each code execution in seconds.
work_dir=temp_dir.name, # Use the temporary directory to store the code files.
)
创建代码生成智能体实例
ConversableAgent class 是对智能体的抽象，autogen中使用该类来创建智能体实例。通常你需要提供一些参数，max_consecutive_auto_reply=1 配置智能体只能回复一次，system_message用来配置系统提示词。llm_config 用来配置大模型作为后端，我选择使用deepseek-chat模型，国产模型性价比非常高，api调用也没有太多限制。代码生成方面，deepseek已经非常优秀了。code_execution_config=False 配置关闭代码执行功能。human_input_mode="NEVER"配置智能体自动回复，不需要人的介入。
以下代码创建了一个负责生成脚本代码的智能体实例。
script_generate_agent = ConversableAgent(
"code_writer_agent",
max_consecutive_auto_reply=1,
system_message=script_executor_message,
llm_config={"config_list": [{"model": "deepseek-chat", "api_key": os.environ["OPENAI_API_KEY"], "base_url":"https://api.deepseek.com"}]},
code_execution_config=False, # Turn off code execution for this agent.
human_input_mode="NEVER",
)
创建代码执行智能体
以下代码创建了代码执行的智能体实例，只负责解析和执行代码，因此该智能体不需要接入大模型，配置llm_config=False。code_execution_config={"executor": script_executor} 用来配置代码执行器。
# Create an agent with code executor configuration.
code_executor_agent = ConversableAgent(
"code_executor_agent",
max_consecutive_auto_reply=1,
system_message=script_executor_message,
llm_config=False,
code_execution_config={"executor": script_executor}, # Use the local command line code executor.
human_input_mode="NEVER",
)初始化任务
以下代码最终将会启动两个智能体，首先脚本生成智能体将根据输入信息生成相应的脚步，代码执行智能体接受脚本生成智能体的输出作为输入，解析脚本，并执行。
chat_result = code_executor_agent.initiate_chat(
script_generate_agent,
message="Get Linux OS information",
)
执行结果如下：
$ python simple_script_executor.py
flaml.automl is not available. Please install flaml[automl] to enable AutoML functionalities.
code_executor_agent (to code_writer_agent):
Get Linux OS information
--------------------------------------------------------------------------------
[autogen.oai.client: 10-07 11:47:55] {349} WARNING - Model deepseek-chat is not found. The cost will be 0. In your config_list, add field {"price" : [prompt_price_per_1k, completion_token_price_per_1k]} for customized pricing.
code_writer_agent (to code_executor_agent):
#```bash 此处因为显示格式的原因加了注释，实际使用时请删除#
#!/bin/bash
# Get Linux OS information
# Display the Linux distribution information
echo "Linux Distribution Information:"
lsb_release -a 2>/dev/null || cat /etc/*release 2>/dev/null || cat /etc/issue 2>/dev/null
echo ""
# Display kernel version
echo "Kernel Version:"
uname -r
echo ""
# Display system architecture
echo "System Architecture:"
uname -m
echo ""
# Display CPU information
echo "CPU Information:"
lscpu
echo ""
# Display memory information
echo "Memory Information:"
free -h
echo ""
# Display disk usage
echo "Disk Usage:"
df -h
echo ""
# Display network interfaces
echo "Network Interfaces:"
ip a
echo ""
# Display hostname
echo "Hostname:"
hostname
echo ""
# Display uptime
echo "Uptime:"
uptime
echo ""
#``` 此处因为显示格式的原因加了注释，实际使用时请删除#
This script will provide a comprehensive overview of your Linux system, including distribution information, kernel version, system architecture, CPU information, memory usage, disk usage, network interfaces, hostname, and uptime.
--------------------------------------------------------------------------------
>>>>>>>> EXECUTING CODE BLOCK (inferred language is bash)...
code_executor_agent (to code_writer_agent):
exitcode: 0 (execution succeeded)
Code output: Linux Distribution Information:
Distributor ID: Ubuntu
Description: Ubuntu 22.04.4 LTS
Release: 22.04
Codename: jammy
Kernel Version:
5.15.153.1-microsoft-standard-WSL2
System Architecture:
x86_64
CPU Information:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Address sizes: 39 bits physical, 48 bits virtual
Byte Order: Little Endian
CPU(s): 12
On-line CPU(s) list: 0-11
Vendor ID: GenuineIntel
Model name: 12th Gen Intel(R) Core(TM) i5-12400
CPU family: 6
Model: 151
Thread(s) per core: 2
Core(s) per socket: 6
Socket(s): 1
Stepping: 2
BogoMIPS: 4991.99
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ss ht syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon rep_good nopl xtopology tsc_reliable nonstop_tsc cpuid pni pclmulqdq vmx ssse3 fma cx16 pdcm sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand hypervisor lahf_lm abm 3dnowprefetch ssbd ibrs ibpb stibp ibrs_enhanced tpr_shadow vnmi ept vpid ept_ad fsgsbase tsc_adjust bmi1 avx2 smep bmi2 erms invpcid rdseed adx smap clflushopt clwb sha_ni xsaveopt xsavec xgetbv1 xsaves avx_vnni umip waitpkg gfni vaes vpclmulqdq rdpid movdiri movdir64b fsrm md_clear serialize arch_lbr flush_l1d arch_capabilities
Virtualization: VT-x
Hypervisor vendor: Microsoft
Virtualization type: full
L1d cache: 288 KiB (6 instances)
L1i cache: 192 KiB (6 instances)
L2 cache: 7.5 MiB (6 instances)
L3 cache: 18 MiB (1 instance)
Vulnerability Gather data sampling: Not affected
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Mmio stale data: Not affected
Vulnerability Retbleed: Mitigation; Enhanced IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling, PBRSB-eIBRS SW sequence
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Memory Information:
total used free shared buff/cache available
Mem: 15Gi 695Mi 14Gi 3.0Mi 591Mi 14Gi
Swap: 4.0Gi 0B 4.0Gi
Disk Usage:
Filesystem Size Used Avail Use% Mounted on
none 7.8G 0 7.8G 0% /usr/lib/modules/5.15.153.1-microsoft-standard-WSL2
none 7.8G 4.0K 7.8G 1% /mnt/wsl
drivers 451G 282G 169G 63% /usr/lib/wsl/drivers
/dev/sdc 251G 24G 215G 10% /
none 7.8G 84K 7.8G 1% /mnt/wslg
none 7.8G 0 7.8G 0% /usr/lib/wsl/lib
rootfs 7.8G 2.2M 7.8G 1% /init
none 7.8G 780K 7.8G 1% /run
none 7.8G 0 7.8G 0% /run/lock
none 7.8G 0 7.8G 0% /run/shm
tmpfs 4.0M 0 4.0M 0% /sys/fs/cgroup
none 7.8G 76K 7.8G 1% /mnt/wslg/versions.txt
none 7.8G 76K 7.8G 1% /mnt/wslg/doc
Network Interfaces:
1: lo: mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
inet 10.25.25.254/32 brd 10.25.25.254 scope global lo
valid_lft forever preferred_lft forever
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: mtu 1500 qdisc mq state UP group default qlen 1000
link/ether 00:15:5d:dd:f0:b7 brd ff:ff:ff:ff:ff:ff
inet 111.21.22.44/20 brd 111.21.22.255 scope global eth0
valid_lft forever preferred_lft forever
inet6 fe80::215:5dff:fedd:f0b7/64 scope link
valid_lft forever preferred_lft forever
Hostname:
PC-XXXXXXYY
Uptime:
11:47:56 up 1 min, 1 user, load average: 0.54, 0.16, 0.05
[…] 智能体实战之autogen：智能体生成shell脚本并执行 […]