🎉 Initial commit

2025-09-03 02:00:16 +00:00 · 2025-02-14 17:56:24 +08:00 · 2025-02-14 17:56:24 +08:00 · bbcd2377a1
commit bbcd2377a1
8 changed files with 752 additions and 0 deletions
--- a/.github/workflows/pypi-publish.yml
+++ b/.github/workflows/pypi-publish.yml
@ -0,0 +1,36 @@
+name: Publish
+
+on:
+  push:
+    tags:
+      - '*'
+  workflow_dispatch:
+
+jobs:
+  pypi-publish:
+    name: Upload release to PyPI
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@master
+    - name: Set up Python
+      uses: actions/setup-python@v1
+      with:
+        python-version: "3.x"
+    - name: Install pypa/build
+      run: >-
+        python -m
+        pip install
+        build
+        --user
+    - name: Build a binary wheel and a source tarball
+      run: >-
+        python -m
+        build
+        --sdist
+        --wheel
+        --outdir dist/
+        .
+    - name: Publish distribution to PyPI
+      uses: pypa/gh-action-pypi-publish@release/v1
+      with:
+        password: ${{ secrets.PYPI_API_TOKEN }}
--- a/.gitignore
+++ b/.gitignore
@ -0,0 +1,176 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+#   Poetry local configuration file - https://python-poetry.org/docs/configuration/#local-configuration
+poetry.toml
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm-python
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# ruff
+.ruff_cache/
+
+# LSP config files
+pyrightconfig.json
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# VisualStudioCode
+.vscode/*
+!.vscode/settings.json
+!.vscode/tasks.json
+!.vscode/launch.json
+!.vscode/extensions.json
+!.vscode/*.code-snippets
--- a/README.md
+++ b/README.md
@ -0,0 +1,157 @@
+<div align="center">
+  <a href="https://v2.nonebot.dev/store"><img src="https://github.com/A-kirami/nonebot-plugin-llmchat/blob/resources/nbp_logo.png" width="180" height="180" alt="NoneBotPluginLogo"></a>
+  <br>
+  <p><img src="https://github.com/A-kirami/nonebot-plugin-llmchat/blob/resources/NoneBotPlugin.svg" width="240" alt="NoneBotPluginText"></p>
+</div>
+
+<div align="center">
+
+# nonebot-plugin-llmchat
+
+_✨ 支持多API预设配置的AI群聊插件 ✨_
+
+
+<a href="./LICENSE">
+    <img src="https://img.shields.io/github/license/FuQuan233/nonebot-plugin-llmchat.svg" alt="license">
+</a>
+<a href="https://pypi.python.org/pypi/nonebot-plugin-llmchat">
+    <img src="https://img.shields.io/pypi/v/nonebot-plugin-llmchat.svg" alt="pypi">
+</a>
+<img src="https://img.shields.io/badge/python-3.9+-blue.svg" alt="python">
+
+</div>
+
+## 📖 介绍
+
+1. **多API预设支持**
+   - 可配置多个LLM服务预设（如不同模型/API密钥）
+   - 支持运行时通过`API预设`命令热切换API配置
+   - 内置服务开关功能（预设名为`off`时停用）
+
+2. **多种回复触发方式**
+   - @触发 + 随机概率触发
+   - 支持处理回复消息
+   - 群聊消息顺序处理，防止消息错乱
+
+3. **分群聊上下文记忆管理**
+   - 分群聊保留对话历史记录（可配置保留条数）
+   - 自动合并未处理消息，降低API用量
+   - 支持`记忆清除`命令手动重置对话上下文
+
+4. **分段回复支持**
+   - 支持多段式回复（由LLM决定如何回复）
+   - 可@群成员（由LLM插入）
+   - 可选输出AI的思维过程（需模型支持）
+
+5. **可自定义性格**
+   - 可动态修改群组专属系统提示词（`/修改设定`）
+   - 支持自定义默认提示词
+
+## 💿 安装
+
+<details open>
+<summary>使用 nb-cli 安装</summary>
+在 nonebot2 项目的根目录下打开命令行, 输入以下指令即可安装
+
+    nb plugin install nonebot-plugin-llmchat
+
+</details>
+
+<details>
+<summary>使用包管理器安装</summary>
+在 nonebot2 项目的插件目录下, 打开命令行, 根据你使用的包管理器, 输入相应的安装命令
+
+<details>
+<summary>pip</summary>
+
+    pip install nonebot-plugin-llmchat
+</details>
+<details>
+<summary>pdm</summary>
+
+    pdm add nonebot-plugin-llmchat
+</details>
+<details>
+<summary>poetry</summary>
+
+    poetry add nonebot-plugin-llmchat
+</details>
+<details>
+<summary>conda</summary>
+
+    conda install nonebot-plugin-llmchat
+</details>
+
+打开 nonebot2 项目根目录下的 `pyproject.toml` 文件, 在 `[tool.nonebot]` 部分追加写入
+
+    plugins = ["nonebot_plugin_llmchat"]
+
+</details>
+
+## ⚙️ 配置
+
+在 nonebot2 项目的`.env`文件中添加下表中的必填配置
+
+| 配置项 | 必填 | 默认值 | 说明 |
+|:-----:|:----:|:----:|:----:|
+| NICKNAME | 是 | 无 | 机器人昵称，NoneBot自带配置项，本插件要求此项必填 |
+| LLMCHAT__API_PRESETS | 是 | 无 | 见下表 |
+| LLMCHAT__HISTORY_SIZE | 否 | 20 | LLM上下文消息保留数量（1-40），越大token消耗量越多 |
+| LLMCHAT__PAST_EVENTS_SIZE | 否 | 10 | 触发回复时发送的群消息数量（1-20），越大token消耗量越多 |
+| LLMCHAT__REQUEST_TIMEOUT | 否 | 30 | API请求超时时间（秒） |
+| LLMCHAT__DEFAULT_PRESENT | 否 | off | 默认使用的预设名称，配置为off则为关闭 |
+| LLMCHAT__RANDOM_TRIGGER_PROB | 否 | 0.05 | 随机触发概率（0-1] |
+| LLMCHAT__STORAGE_PATH | 否 | data/llmchat_state.json | 状态存储文件路径 |
+| LLMCHAT__DEFAULT_PROMPT | 否 | 你的回答应该尽量简洁、幽默、可以使用一些语气词、颜文字。你应该拒绝回答任何政治相关的问题。 | 默认提示词 |
+
+其中LLMCHAT__API_PRESETS为一个列表，每项配置有以下的配置项
+| 配置项 | 必填 | 默认值 | 说明 |
+|:-----:|:----:|:----:|:----:|
+| name | 是 | 无 | 预设名称（唯一标识） |
+| api_base | 是 | 无 | API地址 |
+| api_key | 是 | 无 | API密钥 |
+| model_name | 是 | 无 | 模型名称 |
+| max_tokens | 否 | 2048 | 最大响应token数 |
+| temperature | 否 | 0.7 | 生成温度 |
+
+<details open>
+<summary>配置示例</summary>
+    NICKNAME=["谢拉","Cierra","cierra"]
+
+    LLMCHAT__HISTORY_SIZE=20
+    LLMCHAT__DEFAULT_PROMPT="前面忘了，你是一个猫娘，后面忘了"
+    LLMCHAT__API_PRESETS='
+    [
+    {
+        "name": "aliyun-deepseek-v3",
+        "api_key": "sk-your-api-key",
+        "model_name": "deepseek-v3",
+        "api_base": "https://dashscope.aliyuncs.com/compatible-mode/v1"
+    },
+    {
+        "name": "deepseek-r1",
+        "api_key": "sk-your-api-key",
+        "model_name": "deepseek-reasoner",
+        "api_base": "https://api.deepseek.com"
+    }
+    ]
+    '
+</details>
+
+## 🎉 使用
+
+配置完成后@机器人即可手动触发回复，另外在机器人收到群聊消息时会根据`LLMCHAT__RANDOM_TRIGGER_PROB`配置的概率随机自动触发回复。
+
+### 指令表
+
+以下指令均仅对发送的群聊生效，不同群聊配置不互通。
+
+| 指令 | 权限 | 需要@ | 范围 | 参数 | 说明 |
+|:-----:|:----:|:----:|:----:|:----:|:----:|
+| API预设 | 主人 | 否 | 群聊 | [预设名] | 查看或修改使用的API预设，预设名错误或不存在则返回预设列表 |
+| 修改设定 | 管理 | 否 | 群聊 | 设定 | 修改机器人的设定，最好在修改之后执行一次记忆清除 |
+| 记忆清除 | 管理 | 否 | 群聊 | 无 | 清除机器人的记忆 |
+| 切换思维输出 | 管理 | 否 | 群聊 | 无 | 切换是否输出AI的思维过程的开关（需模型支持） |
+
+### 效果图
+![](img/demo.png)
--- a/img/demo.png
+++ b/img/demo.png
--- a/nonebot_plugin_llmchat/init.py
+++ b/nonebot_plugin_llmchat/init.py
@ -0,0 +1,334 @@
+import aiofiles
+from nonebot import get_plugin_config, on_message, logger, on_command, get_driver
+from nonebot.plugin import PluginMetadata
+from nonebot.adapters.onebot.v11 import GroupMessageEvent, Message
+from nonebot.adapters.onebot.v11.permission import GROUP_ADMIN, GROUP_OWNER
+from nonebot.params import CommandArg
+from nonebot.rule import Rule
+from nonebot.permission import SUPERUSER
+from typing import Dict
+from datetime import datetime
+from collections import deque
+import asyncio
+from openai import AsyncOpenAI
+from .config import Config, PresetConfig
+import time
+import json
+import os
+import random
+from apscheduler.schedulers.asyncio import AsyncIOScheduler
+import time
+
+__plugin_meta__ = PluginMetadata(
+    name="llmchat",
+    description="支持多API预设配置的AI群聊插件",
+    usage="""@机器人 + 消息 开启对话""",
+    type="application",
+    homepage="https://github.com/FuQuan233/nonebot-plugin-llmchat",
+    config=Config,
+    supported_adapters={"~onebot.v11"},
+)
+
+pluginConfig = get_plugin_config(Config).llmchat
+driver = get_driver()
+
+# 初始化群组状态
+class GroupState:
+    def __init__(self):
+        self.preset_name = pluginConfig.default_preset
+        self.history = deque(maxlen=pluginConfig.history_size)
+        self.queue = asyncio.Queue()
+        self.processing = False
+        self.last_active = time.time()
+        self.past_events = deque(maxlen=pluginConfig.past_events_size)
+        self.group_prompt = None
+        self.output_reasoning_content = False
+
+group_states: Dict[int, GroupState] = {}
+
+# 获取当前预设配置
+def get_preset(group_id: int) -> PresetConfig:
+    state = group_states[group_id]
+    for preset in pluginConfig.api_presets:
+        if preset.name == state.preset_name:
+            return preset
+    return pluginConfig.api_presets[0]  # 默认返回第一个预设
+
+# 消息格式转换
+def format_message(event: GroupMessageEvent) -> Dict:
+    text_message = ""
+    if event.reply != None:
+        text_message += f"[回复 {event.reply.sender.nickname} 的消息 {event.reply.message.extract_plain_text()}]\n"
+
+    if event.is_tome():
+        text_message += f"@{list(driver.config.nickname)[0]} "
+
+    for msgseg in event.get_message():
+        if msgseg.type == "at":
+            text_message += msgseg.data.get("name", "")
+        elif msgseg.type == "image":
+            text_message += "[图片]"
+        elif msgseg.type == "voice":
+            text_message += "[语音]"
+        elif msgseg.type == "face":
+            pass
+        elif msgseg.type == "text":
+            text_message += msgseg.data.get("text", "")
+
+    message =  {
+        "SenderNickname": str(event.sender.card or event.sender.nickname),
+        "SenderUserId": str(event.user_id),
+        "Message": text_message,
+        "SendTime" : datetime.fromtimestamp(event.time).isoformat()
+    }
+    return json.dumps(message, ensure_ascii=False)
+
+async def isTriggered(event: GroupMessageEvent) -> bool:
+    """扩展后的消息处理规则"""
+    
+    group_id = event.group_id
+
+    if group_id not in group_states:
+        logger.info(f"初始化群组状态，群号：{group_id}")
+        group_states[group_id] = GroupState()
+    
+    state = group_states[group_id]
+
+    if state.preset_name == "off":
+        return False
+
+    state.past_events.append(event)
+
+    # 原有@触发条件
+    if event.is_tome():
+        return True
+    
+    # 随机触发条件
+    if random.random() < pluginConfig.random_trigger_prob:
+        return True
+    
+    return False
+
+
+# 消息处理器
+handler = on_message(
+    rule=Rule(isTriggered),
+    priority=10,
+    block=False,
+)
+
+@handler.handle()
+async def handle_message(event: GroupMessageEvent):
+    group_id = event.group_id
+    logger.debug(f"收到群聊消息 群号：{group_id} 用户：{event.user_id} 内容：{event.get_plaintext()}")
+
+    if group_id not in group_states:
+        group_states[group_id] = GroupState()
+    
+    state = group_states[group_id]
+    
+    await state.queue.put(event)
+    if not state.processing:
+        state.processing = True
+        asyncio.create_task(process_messages(group_id))
+
+async def process_messages(group_id: int):
+    state = group_states[group_id]
+    preset = get_preset(group_id)
+    
+    # 初始化OpenAI客户端
+    client = AsyncOpenAI(
+        base_url=preset.api_base,
+        api_key=preset.api_key,
+        timeout=pluginConfig.request_timeout
+    )
+    
+    logger.info(f"开始处理群聊消息 群号：{group_id} 当前队列长度：{state.queue.qsize()}")
+    while not state.queue.empty():
+        event = await state.queue.get()
+        logger.debug(f"从队列获取消息 群号：{group_id} 消息ID：{event.message_id}")
+        try:
+            systemPrompt = (
+f'''
+我想要你帮我在群聊中闲聊，大家一般叫你{"、".join(list(driver.config.nickname))}，我将会在后面的信息中告诉你每条群聊信息的发送者和发送时间，你可以直接称呼发送者为他对应的昵称。
+你的回复需要遵守以下几点规则：
+- 你可以使用多条消息回复，每两条消息之间使用<botbr>分隔，<botbr>前后不需要包含额外的换行和空格。
+- 除<botbr>外，消息中不应该包含其他类似的标记。
+- 不要使用markdown格式，聊天软件不支持markdown解析。
+- 你应该以普通人的方式发送消息，每条消息字数要尽量少一些，应该倾向于使用更多条的消息回复。
+- 代码则不需要分段，用单独的一条消息发送。
+- 请使用发送者的昵称称呼发送者，你可以礼貌地问候发送者，但只需要在第一次回答这位发送者的问题时问候他。
+- 你有at群成员的能力，只需要在某条消息中插入[CQ:at,qq=（QQ号）]，也就是CQ码。at发送者是非必要的，你可以根据你自己的想法at某个人。
+- 如果有多条消息，你应该优先回复提到你的，一段时间之前的就不要回复了，也可以直接选择不回复。
+- 如果你需要思考的话，你应该思考尽量少，以节省时间。
+下面是关于你性格的设定，如果设定中提到让你扮演某个人，或者设定中有提到名字，则优先使用设定中的名字。
+{state.group_prompt or pluginConfig.default_prompt}
+'''
+            )
+
+            messages = [{"role": "system", "content": systemPrompt}]
+
+            messages += list(state.history)[-pluginConfig.history_size:]
+
+            # 没有未处理的消息说明已经被处理了，跳过
+            if state.past_events.__len__() < 1:
+                break
+
+            # 将机器人错过的消息推送给LLM
+            content = ",".join([format_message(ev) for ev in state.past_events])
+
+            logger.debug(f"发送API请求 模型：{preset.model_name} 历史消息数：{len(messages)}")
+            response = await client.chat.completions.create(
+                model=preset.model_name,
+                messages=messages + [{"role": "user", "content": content}],
+                max_tokens=preset.max_tokens,
+                temperature=preset.temperature,
+                timeout=60
+            )
+            logger.debug(f"收到API响应 使用token数：{response.usage.total_tokens}")
+
+            reply = response.choices[0].message.content
+
+            # 请求成功后再保存历史记录，保证user和assistant穿插，防止R1模型报错
+            state.history.append({"role": "user", "content": content})
+            state.past_events.clear()
+
+            reasoning_content: str | None = getattr(response.choices[0].message, "reasoning_content", None)
+            if state.output_reasoning_content and reasoning_content:
+                await handler.send(Message(reasoning_content))
+
+            logger.info(f"准备发送回复消息 群号：{group_id} 消息分段数：{len(reply.split('<botbr>'))}")
+            for r in reply.split("<botbr>"):
+                # 删除前后多余的换行和空格
+                while r[0] == "\n" or r[0] == " ": r = r[1:]
+                while r[-1] == "\n" or r[0] == " ": r = r[:-1]
+                await asyncio.sleep(2)
+                logger.debug(f"发送消息分段 内容：{r[:50]}...")  # 只记录前50个字符避免日志过大
+                await handler.send(Message(r))
+            
+            # 添加助手回复到历史
+            state.history.append({
+                "role": "assistant",
+                "content": reply,
+            })
+            
+        except Exception as e:
+            logger.error(f"API请求失败 群号：{group_id} 错误：{str(e)}", exc_info=True)
+            await handler.send(Message(f"服务暂时不可用，请稍后再试\n{str(e)}"))
+        finally:
+            state.queue.task_done()
+    
+    state.processing = False
+
+# 预设切换命令
+preset_handler = on_command("API预设", priority=1, block=True, permission=SUPERUSER)
+@preset_handler.handle()
+async def handle_preset(event: GroupMessageEvent, args: Message = CommandArg()):
+    group_id = event.group_id
+    preset_name = args.extract_plain_text().strip()
+    
+    if group_id not in group_states:
+        group_states[group_id] = GroupState()
+    
+    if preset_name == "off":
+        group_states[group_id].preset_name = preset_name
+        await preset_handler.finish(f"已关闭llmchat")
+
+    available_presets = {p.name for p in pluginConfig.api_presets}
+    if preset_name not in available_presets:
+        await preset_handler.finish(f"当前API预设：{group_states[group_id].preset_name}\n可用API预设：\n- {'\n- '.join(available_presets)}")
+    
+    group_states[group_id].preset_name = preset_name
+    await preset_handler.finish(f"已切换至API预设：{preset_name}")
+
+preset_handler = on_command("修改设定", priority=1, block=True, permission=(SUPERUSER|GROUP_ADMIN|GROUP_OWNER))
+@preset_handler.handle()
+async def handle_preset(event: GroupMessageEvent, args: Message = CommandArg()):
+    group_id = event.group_id
+    group_prompt = args.extract_plain_text().strip()
+    
+    if group_id not in group_states:
+        group_states[group_id] = GroupState()
+    
+    group_states[group_id].group_prompt = group_prompt
+    await preset_handler.finish("修改成功")
+
+reset_handler = on_command("记忆清除", priority=99, block=True, permission=(SUPERUSER|GROUP_ADMIN|GROUP_OWNER))
+@reset_handler.handle()
+async def handle_reset(event: GroupMessageEvent, args: Message = CommandArg()):
+    group_id = event.group_id
+
+    if group_id not in group_states:
+        group_states[group_id] = GroupState()
+
+    group_states[group_id].past_events.clear()
+    group_states[group_id].history.clear()
+    await preset_handler.finish(f"记忆已清空")
+
+# 预设切换命令
+preset_handler = on_command("切换思维输出", priority=1, block=True, permission=(SUPERUSER|GROUP_ADMIN|GROUP_OWNER))
+@preset_handler.handle()
+async def handle_preset(event: GroupMessageEvent, args: Message = CommandArg()):
+    group_id = event.group_id
+    
+    if group_id not in group_states:
+        group_states[group_id] = GroupState()
+    
+    if group_states[group_id].output_reasoning_content:
+        group_states[group_id].output_reasoning_content = False
+        await preset_handler.finish("已关闭思维输出")
+    else:
+        group_states[group_id].output_reasoning_content = True
+        await preset_handler.finish("已开启思维输出")
+
+
+# region 持久化与定时任务
+async def save_state():
+    """保存群组状态到文件"""
+    logger.info(f"开始保存群组状态到文件：{pluginConfig.storage_path}")
+    data = {
+        gid: {
+            "preset": state.preset_name,
+            "history": list(state.history),
+            "last_active": state.last_active,
+            "group_prompt": state.group_prompt,
+            "output_reasoning_content": state.output_reasoning_content
+        }
+        for gid, state in group_states.items()
+    }
+    
+    os.makedirs(os.path.dirname(pluginConfig.storage_path), exist_ok=True)
+    async with aiofiles.open(pluginConfig.storage_path, "w") as f:
+        await f.write(json.dumps(data, ensure_ascii=False))
+
+async def load_state():
+    """从文件加载群组状态"""
+    logger.info(f"从文件加载群组状态：{pluginConfig.storage_path}")
+    if not os.path.exists(pluginConfig.storage_path):
+        return
+    
+    async with aiofiles.open(pluginConfig.storage_path, "r") as f:
+        data = json.loads(await f.read())
+        for gid, state_data in data.items():
+            state = GroupState()
+            state.preset_name = state_data["preset"]
+            state.history = deque(state_data["history"], maxlen=pluginConfig.history_size)
+            state.last_active = state_data["last_active"]
+            state.group_prompt = state_data["group_prompt"]
+            state.output_reasoning_content = state_data["output_reasoning_content"]
+            group_states[int(gid)] = state
+
+# 注册生命周期事件
+@driver.on_startup
+async def init_plugin():
+    logger.info("插件启动初始化")
+    await load_state()
+    scheduler = AsyncIOScheduler()
+    # 每5分钟保存状态
+    scheduler.add_job(save_state, 'interval', minutes=5)
+    scheduler.start()
+
+@driver.on_shutdown
+async def cleanup_plugin():
+    logger.info("插件关闭清理")
+    await save_state()
--- a/nonebot_plugin_llmchat/config.py
+++ b/nonebot_plugin_llmchat/config.py
@ -0,0 +1,25 @@
+from pydantic import BaseModel, Field
+from typing import List, Dict, Optional
+
+class PresetConfig(BaseModel):
+    """API预设配置"""
+    name: str = Field(..., description="预设名称（唯一标识）")
+    api_base: str = Field(..., description="API基础地址")
+    api_key: str = Field(..., description="API密钥")
+    model_name: str = Field(..., description="模型名称")
+    max_tokens: int = Field(2048, description="最大响应token数")
+    temperature: float = Field(0.7, description="生成温度（0-2]")
+
+class ScopedConfig(BaseModel):
+    """LLM Chat Plugin配置"""
+    api_presets: List[PresetConfig] = Field(...,description="API预设列表（至少配置1个预设）")
+    history_size: int = Field(20, description="LLM上下文消息保留数量")
+    past_events_size : int = Field(10, description="触发回复时发送的群消息数量")
+    request_timeout: int = Field(30, description="API请求超时时间（秒）")
+    default_preset: str = Field("off", description="默认使用的预设名称")
+    random_trigger_prob: float = Field(0.05, ge=0.0, le=1.0, description="随机触发概率（0-1]")
+    storage_path: str = Field("data/llmchat_state.json", description="状态存储文件路径")
+    default_prompt: str = Field("你的回答应该尽量简洁、幽默、可以使用一些语气词、颜文字。你应该拒绝回答任何政治相关的问题。", description="默认提示词")
+
+class Config(BaseModel):
+    llmchat: ScopedConfig
--- a/nonebot_plugin_llmchat/nonebot_plugin_llmchat
+++ b/nonebot_plugin_llmchat/nonebot_plugin_llmchat
@ -0,0 +1 @@
+nonebot_plugin_llmchat
--- a/pyproject.toml
+++ b/pyproject.toml
@ -0,0 +1,23 @@
+[tool.poetry]
+name = "nonebot-plugin-llmchat"
+version = "0.1.0"
+description = "Nonebot AI group chat plugin supporting multiple API preset configurations"
+license = "GPL"
+authors = ["FuQuan i@fuquan.moe"]
+readme = "README.md"
+homepage = "https://github.com/FuQuan233/nonebot-plugin-llmchat"
+repository = "https://github.com/FuQuan233/nonebot-plugin-llmchat"
+documentation = "https://github.com/FuQuan233/nonebot-plugin-llmchat#readme"
+keywords = ["nonebot", "nonebot2", "llm", "ai"]
+
+[tool.poetry.dependencies]
+python = "^3.9"
+openai = ">=1.0.0>"
+nonebot2 = "^2.2.0"
+aiofiles = ">=24.0.0"
+nonebot-plugin-apscheduler = "^0.5.0"
+nonebot-adapter-onebot = "^2.0.0"
+
+[build-system]
+requires = ["poetry-core>=1.0.0"]
+build-backend = "poetry.core.masonry.api"