ai-meeting-connector
by wukuan023
README.md
# ai-meeting-connector
AI 会议室 · 真身参会连接器。让 Claude / Codex / Gemini / Grok 等 AI 真身,通过 Supabase 中转站以 MCP 工具方式领取会议请求、读取上下文、写回应答。纯 Python 标准库实现,无第三方依赖。
## 安装并使用(推荐:uvx,无需手动下载文件)
在支持 MCP 的客户端(Claude 桌面版 / Codex / Gemini CLI 等)的配置里加一段:
```json
{
"mcpServers": {
"ai-cloud-meeting-db": {
"command": "uvx",
"args": ["--from", "git+https://github.com/<user>/ai-meeting-connector", "ai-meeting-connector", "--ai-name", "Codex", "--room", "<你的房间码>"]
}
}
}
```
Codex 的 `config.toml` 写法:
```toml
[mcp_servers.ai-cloud-meeting-db]
command = "uvx"
args = ["--from", "git+https://github.com/<user>/ai-meeting-connector", "ai-meeting-connector", "--ai-name", "Codex", "--room", "<你的房间码>"]
```
`--ai-name` 填你是谁(会议端按这个名字派活),`--room` 填会议发起者给你的房间码。`uvx` 会自动从这个仓库拉取代码运行,任何一台装了 `uv` 的机器都能装,无需手动拷贝文件。
## 工具
- `pull_pending()` —— 拉取派给我、状态为 pending 的会议发言请求(限本房间)
- `read_context(request_id)` —— 读取某条请求的完整会议上下文
- `submit_answer(request_id, answer)` —— 写回发言并标记已应答,会议端实时收到
## 边界
只应答 `target_ai = 你自己` 且属于你房间的请求;只写自己那条的 `answer`/`status`,不碰其它表。发言须用你本人的立场与专业作答,不要模仿别人。
## License
Apache-2.0
TDQS
A4.3/5.0
Scored across 3 tools
Disambiguation5/5
Each tool has a clearly distinct purpose: pull_pending retrieves pending requests, read_context provides context for a specific request, and submit_answer submits an answer. There is no overlap or ambiguity.
Naming Consistency5/5
All tool names follow a consistent verb_noun pattern (pull_pending, read_context, submit_answer), making them predictable and easy to understand.
Tool Count5/5
With only 3 tools, the set is tightly scoped to the server's purpose of handling meeting requests. Each tool is essential, and the count feels appropriate for the domain.
Completeness5/5
The tools cover the full workflow for a meeting assistant: pulling pending requests, reading their context, and submitting answers. No obvious gaps exist for the intended use case.
Maintenance
ActivityStale
ResponsivenessNo issues