Cognigy Ai Mcp Management Server

io.github.TsvetanG2由 GitHub 账号 TsvetanG2 发布v0.2.1更新于 Oct 24, 2026

MCP server for Cognigy.AI - 132 tools to build, configure & operate conversational AI agents

概览

AI 生成的概览

本地 MCP 服务器,通过 132 个 Management API 工具让助手构建、配置、测试和运营 Cognigy.AI 对话式 AI 智能体。

可以做什么
管理项目、流程、节点、意图和 NLU 训练,包括评分与审计检查。运行 playbook 和回归测试,注入或重置会话上下文。创建、恢复、对比和提升快照,创建、合并、上传或下载包。配置 LLM、NLU 连接器、Knowledge AI 存储与来源、函数和扩展。处理联系人档案、分析、审计日志、人工坐席转接和组织级搜索。跟踪长时间运行的异步任务,并对列表结果分页。
适用场景
适合在 Cognigy.AI 上开发的开发者、解决方案架构师和合作伙伴,希望用 AI 助手代替 Cognigy 界面来脚本化修改智能体、运行 NLU 训练与回归测试,或在环境间提升快照和包。
运行要求
需要 Node.js 20 或更高版本,并单独安装 @cognigy/rest-api-client 对等依赖。需要具备 API 访问权限的 Cognigy.AI 账户(试用、SaaS 或本地部署)。必须设置环境变量 COGNIGY_BASE_URL 和 COGNIGY_API_KEY,COGNIGY_DEFAULT_PROJECT_ID 可选。需要能访问 Cognigy API 端点的网络。
安装前请注意
需要在 COGNIGY_API_KEY 中提供 Cognigy API 密钥,请勿写入提交的配置文件。变更类工具默认 dryRun 为 true,但启用后可实际修改流程、快照、包、档案和会话。对等依赖采用 Cognigy 专有许可,安装即表示接受。联系人档案工具处理个人数据,包括合并和 GDPR 导出。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Cognigy Ai Mcp Management Server,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

Cognigy.AI Management MCP Server

[CI] [npm version] [npm downloads] [node] [License: MIT] [Glama Score] [MCP Registry] [MCP Badge]

[cognigy-ai-mcp-management-server]

Model Context Protocol server for managing Cognigy.AI virtual agents through the Management API

This is an independent, open-source MCP server and is not affiliated with, endorsed by, or sponsored by Cognigy or NiCE. It requires your own valid Cognigy.AI account and API key, used in accordance with Cognigy's commercial license terms. "Cognigy" and "Cognigy.AI" are trademarks of their respective owners.

What is this?

cognigy-ai-mcp-management-server is a local MCP server that enables AI coding assistants (Claude, Cursor, etc.) to build, configure, test, and operate Cognigy.AI conversational AI agents programmatically.

Instead of clicking through the Cognigy UI, you can now:

  • Create and manage flows, nodes, and intents via natural language
  • Run NLU training and regression tests automatically
  • Deploy snapshots and packages across environments
  • Configure LLMs, Knowledge AI, and integrations

Built for: Developers, solution architects, and SI partners who build on Cognigy.AI.

Requirements

  • Node.js 20.0.0 or higher
  • Cognigy.AI account with API access (Trial, SaaS, or on-premises)
  • API Key generated from Cognigy UI (My Profile → API Keys)

Installation

bash
# 1. Install the Cognigy REST API client (required peer dependency)npm install @cognigy/rest-api-client
# 2. Install this MCP servernpm install -g cognigy-ai-mcp-management-server

Important: The @cognigy/rest-api-client package is licensed under "Cognigy Proprietary License". By installing it, you agree to Cognigy's license terms.

Configuration

Set environment variables before running:

bash
# Required: Your Cognigy API endpointexport COGNIGY_BASE_URL=https://api-trial.cognigy.ai
# Required: Your API key (never commit this!)export COGNIGY_API_KEY=your-api-key-here
# Optional: Default project for operationsexport COGNIGY_DEFAULT_PROJECT_ID=your-project-id

Common Base URLs:

EnvironmentBase URL
Trialhttps://api-trial.cognigy.ai
SaaS EUhttps://api-app.cognigy.ai
SaaS UShttps://api-app-us.cognigy.ai
Dedicatedhttps://api-<company>.cognigy.cloud

MCP Client Configuration

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

json
{  "mcpServers": {    "cognigy": {      "command": "npx",      "args": ["cognigy-ai-mcp-management-server"],      "env": {        "COGNIGY_BASE_URL": "https://api-trial.cognigy.ai",        "COGNIGY_API_KEY": "your-api-key-here"      }    }  }}

Claude Code

Add to .mcp.json in your project root:

json
{  "mcpServers": {    "cognigy": {      "command": "npx",      "args": ["cognigy-ai-mcp-management-server"],      "env": {        "COGNIGY_BASE_URL": "https://api-trial.cognigy.ai",        "COGNIGY_API_KEY": "your-api-key-here"      }    }  }}

Cursor

Add to .cursor/mcp.json:

json
{  "mcpServers": {    "cognigy": {      "command": "npx",      "args": ["cognigy-ai-mcp-management-server"],      "env": {        "COGNIGY_BASE_URL": "https://api-trial.cognigy.ai",        "COGNIGY_API_KEY": "your-api-key-here"      }    }  }}

Available Tools

All Tools

This server provides 132 MCP tools organized by domain:

CategoryToolsDescription
Projects & Flows5List projects, flows, flow settings, logs
Nodes9CRUD operations, search, move, AI output generation
Intents & NLU10Intent management, training, scoring, audit
Endpoints2List and inspect endpoint configurations
Sessions2Inject/reset conversation context
Conversations3Fetch conversations, transcripts
Playbooks & Testing6Run playbooks, regression tests
Snapshots9Create, restore, package, diff, promote
Packages7Create, merge, upload, download
Connections5Manage API connections (secrets redacted)
LLMs7Configure generative AI providers
NLU Connectors5External NLU (Dialogflow, LUIS, etc.)
Knowledge AI21Stores, sources, chunks, connectors (RAG)
Functions9Custom code functions, instances
Extensions6Upload, configure, manage extensions
Contact Profiles11Profile CRUD, merge, export (GDPR)
Analytics4Conversation, call, knowledge metrics
Audit2Audit event logs
Handover7Live agent providers and services
Search1Organization-wide resource search
Tasks2Async task status tracking

Key Features

  • Safe by default: All mutating tools use dryRun: true by default
  • Async-aware: Long-running operations poll until completion
  • Security: API keys only in memory, secrets automatically redacted
  • Pagination: All list operations support limit and skip

Development & Testing

Live API Testing

bash
# Create .env from templatecp .env.example .env# Edit .env with your credentials
# Run development servernpm run dev

Build

bash
npm run build    # Compile TypeScriptnpm test         # Run test suite (49 tests)npm run lint     # Check code style

Regenerate Types (optional)

If you need to update the generated types from a newer Cognigy API:

bash
npm run update:spec   # Download latest OpenAPI specnpm run gen:types     # Regenerate TypeScript types

Important: Dependencies

This package requires @cognigy/rest-api-client as a peer dependency. This means:

  1. You must install it separately (see Installation above)
  2. You accept Cognigy's license terms by installing their package
  3. We do not bundle or redistribute any Cognigy code

You are responsible for:

  • Reviewing and accepting Cognigy's proprietary license terms
  • Having appropriate rights to use the Cognigy API
  • Keeping your API credentials secure

The Cognigy REST API client is the official SDK maintained by Cognigy GmbH and is published on npm under "Cognigy Proprietary License".

License

This MCP server is released under the MIT License. See LICENSE for details.

Note: The MIT license applies only to this MCP server code. Dependencies (particularly @cognigy/rest-api-client) are subject to their own licenses.

Contributing

Contributions are welcome! Please:

  1. Fork the repository
  2. Create a feature branch (git checkout -b feature/amazing-feature)
  3. Follow existing code patterns (Zod schemas, dryRun flags, etc.)
  4. Add tests for new tools
  5. Run npm test && npm run build before committing
  6. Submit a Pull Request

Development Guidelines

  • All tools must have Zod input validation
  • Mutating tools must support dryRun flag (default: true)
  • Tool descriptions should be LLM-friendly (1-2 sentences)
  • Never log or expose API keys/secrets
  • Test against mock server before live API

Support


Built with Model Context Protocol SDK

来源:README.md,提交 e647397

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

3
  1. v0.2.1最新Oct 5, 2026
  2. v0.2.0Oct 4, 2026
  3. v0.1.4Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分