DebateTalk MCP
Enables AI assistants to facilitate structured multi-model debates that synthesize multiple perspectives into clear categories like ground truths and blind spots. It provides tools for running real-time debates, checking model health, and managing history via the Model Context Protocol.
README
DebateTalk MCP
Official MCP server and CLI for DebateTalk — run structured multi-model AI debates from your AI assistant or terminal.
DebateTalk makes multiple AI models argue a question independently, challenge each other's reasoning, and converge on a structured synthesis: Strong Ground, Fault Lines, Blind Spots, and Your Call.
Features
- MCP server — connect Claude Desktop, Cursor, or any MCP-compatible client to DebateTalk
- CLI — run debates and check model status from the terminal
- Streaming output — debates stream in real time via SSE
- 5 tools:
run_debate,get_model_status,recommend_models,estimate_cost,get_history
Quickstart
Claude Code — plugin marketplace
1. Add the DebateTalk marketplace:
/plugin marketplace add DebateTalk-AI/mcp
2. Install the plugin:
/plugin install debatetalk@debatetalk-mcp
3. Set your API key:
Get a key at console.debatetalk.ai/api-keys, then add it to ~/.claude/settings.json:
{
"pluginConfigs": {
"debatetalk@debatetalk-mcp": {
"options": {
"api_key": "dt_your_key_here"
}
}
}
}
Then run /reload-plugins — the five DebateTalk tools are immediately available in your session.
MCP (Claude Desktop, Cursor, Cline, Goose, and any MCP-compatible client)
1. Get an API key
Create a key at console.debatetalk.ai/api-keys. Requires a Pro or Enterprise plan. Free tier: 5 debates/day.
2. Add to your MCP client config
{
"mcpServers": {
"dt": {
"command": "npx",
"args": ["-y", "@debatetalk/mcp"],
"env": {
"DEBATETALK_API_KEY": "dt_your_key_here"
}
}
}
}
Config file locations:
- Claude Desktop (Mac):
~/Library/Application Support/Claude/claude_desktop_config.json - Claude Desktop (Windows):
%APPDATA%\Claude\claude_desktop_config.json - Claude Code:
~/.claude/settings.json(undermcpServers) - Cursor:
.cursor/mcp.jsonin your project root - Windsurf:
~/.codeium/windsurf/mcp_config.json - Cline / Roo Code: MCP settings panel in VS Code extension
- Goose:
~/.config/goose/config.yaml(underextensions) - Other clients: refer to your client's MCP documentation
3. Ask your AI assistant to run a debate
MCP clients read the tool description to decide when to call it — no exact phrasing required. Any of these work:
"debate whether we should rewrite our backend in Go" "use DT — should we raise our Series A now?" "multi-model this: is Rust worth learning in 2026?" "stress-test this architecture decision" "get a second opinion on moving to microservices"
Claude will also invoke it proactively for high-stakes decisions where a single AI answer is insufficient.
CLI
Install globally:
npm install -g @debatetalk/mcp
Set your API key:
export DEBATETALK_API_KEY=dt_your_key_here
Run a debate:
dt debate "Should we adopt microservices?"
Check which models are online:
dt models
Get a recommended model panel for your question:
dt recommend "Is Rust worth learning in 2026?"
Estimate cost before running:
dt cost "Should we raise our Series A now?"
View past debates:
dt history
dt history --limit 5
MCP Tools Reference
| Tool | Auth required | Description |
|---|---|---|
run_debate |
Yes | Run a structured multi-model debate (streaming) |
get_model_status |
No | Real-time health and latency for all models |
recommend_models |
No | Get the best model panel for your question |
estimate_cost |
Yes | Estimate credit cost before running |
get_history |
Yes | List your past debates |
run_debate
question string required The question or topic to debate
models array optional Specific model IDs to use (omit for smart routing)
rounds number optional Number of deliberation rounds (default: 2)
get_model_status
No parameters. Returns live health, latency, and uptime per model.
recommend_models
question string required The question — routing picks the strongest panel
estimate_cost
question string required
models array optional
rounds number optional
get_history
limit number optional Number of debates to return (default: 20, max: 100)
Configuration
| Variable | Required | Description |
|---|---|---|
DEBATETALK_API_KEY |
For authenticated tools | Your API key from console.debatetalk.ai |
Public tools (get_model_status, recommend_models) work without an API key.
Plans & Limits
| Plan | Debates/day | API keys | Debaters |
|---|---|---|---|
| Free | 5 | — | 3 |
| Pro | Unlimited | 2 | 5 |
| Enterprise | Unlimited | Unlimited | 10 |
Development
git clone https://github.com/DebateTalk-AI/mcp
cd mcp
npm install
npm run build
npm test
Run MCP server locally:
DEBATETALK_API_KEY=dt_your_key npm run dev:mcp
Run CLI locally:
DEBATETALK_API_KEY=dt_your_key npm run dev:cli -- debate "your question"
Contributing
See CONTRIBUTING.md. Issues and PRs welcome.
License
MIT — see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。