topolift-mcp
Exposes TopoLift's atom-grounded negotiation reasoning to MCP-capable agents, providing tools to fetch a negotiation vocabulary and perform structured negotiation with traceable citations.
README
topolift-mcp
MCP server exposing TopoLift's atom-grounded negotiation reasoning to any MCP-capable agent — Claude Code, Cursor, OpenAI Agents SDK, etc.
Listed on the official MCP Registry as
io.github.TopoLift/topolift-mcp.mcp-name: io.github.TopoLift/topolift-mcp
What you get
Two tools:
topolift_dialect— fetches TopoLift's published vocabulary (regimes, canonical strategies, signal keys, citation grammar). Call this once at session start; no auth required.topolift_negotiate— sends a negotiation context to TopoLift's reasoning engine and returns a bilingual response:- A typed
topologyslot (regime,load_bearing_strategies[],bridge_pivots[],topology_signals{}) drawn from a closed vocabulary - Prose fields with inline
[Cluster_X#strategy1,strategy2]citation tokens anchoring every claim to the cluster + load-bearing atoms that drove it
- A typed
Atoms stay on the TopoLift server. What travels is the grammar of the dialect — the vocabulary your agent uses to read structural reasoning.
Install
pip install topolift-mcp
Set your API key (get one at https://topolift.ai):
export TOPOLIFT_API_KEY=tl-...
Claude Code
claude mcp add topolift-negotiation -- topolift-mcp
…or, with the API key inline:
claude mcp add topolift-negotiation -e TOPOLIFT_API_KEY=tl-... -- topolift-mcp
Cursor / Continue / other MCP clients
Add to your MCP config (~/.cursor/mcp.json or equivalent):
{
"mcpServers": {
"topolift-negotiation": {
"command": "topolift-mcp",
"env": {
"TOPOLIFT_API_KEY": "tl-..."
}
}
}
}
Running directly (no install)
TOPOLIFT_API_KEY=tl-... python -m topolift_mcp.server
Configuration
Environment variables:
| Var | Default | Purpose |
|---|---|---|
TOPOLIFT_API_KEY |
(required for negotiate) | Bearer key; get one at https://topolift.ai |
TOPOLIFT_API_URL |
https://api.topolift.ai |
API base URL |
TOPOLIFT_TIMEOUT |
600 |
Negotiate-call timeout in seconds |
TOPOLIFT_LOG_LEVEL |
INFO |
Python logging level |
The dialect tool works without TOPOLIFT_API_KEY — only topolift_negotiate requires it.
Pricing
- Bearer key (this MCP server's path): one-time / monthly plans starting at $50. See https://topolift.ai.
- x402 micropayment: agents can pay $0.10 USDC per call directly against
https://api.topolift.ai/v1/negotiatewith no API key — see the API's 402 challenge for details.
How it works
- Your agent calls
topolift_dialectonce and learns the vocabulary. - Your agent calls
topolift_negotiatewith a scenario. - TopoLift retrieves the most relevant atom clusters from its server-side topology, runs a Mistral-Small-4-119B reasoning pass with a closed-vocabulary prompt, and returns the bilingual response.
- Your agent parses
topologyfor machine-readable structure and reads the prose fields with citation-traceable evidence.
The atoms — the structural primitives — never leave TopoLift's servers. The dialect — the names and the grammar — is published openly so any agent can read structural reasoning fluently.
License
MIT
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。