COR MCP Server

COR MCP Server

Comprehensive MCP server for Project COR, a project management platform, exposing 61 tools covering the complete COR REST API as MCP tools.

Category
访问服务器

README

COR MCP Server

Comprehensive MCP server for Project COR — a project management platform for creative and professional teams.

Exposes 61 tools covering the complete COR REST API surface as MCP tools, usable from any MCP-compatible client (Hermes Agent, Claude Desktop, Cursor, Claude Code, Cline, Windsurf, etc.).

Features

  • 61 MCP tools — full CRUD for all COR entities across 11 modules
  • Dual authentication — pick what suits you:
    • User Credentials (email + password) — quick personal use
    • Client Credentials (API key + client secret) — server-to-server
  • Async everywhere — httpx + asyncio, non-blocking I/O
  • Dual transport — stdio (default for agents) and sse (HTTP for remote access)
  • OAuth token auto-refresh — never worry about expiry mid-session
  • Docker support — one-command deployment
  • Hermes Agent skill included — skills/cor-mcp-setup/ for easy integration
  • 24 passing tests — fully mocked, no live credentials required

Entity Coverage

Module Tools Covers
Projects 14 CRUD, collaborators, costs, labels, ratecards, templates, profitability
Tasks 10 Search, CRUD, collaborators, labels
Clients 6 CRUD, fees
Contracts 5 CRUD, positions
Time Tracking 5 Log, search, status, accept suggested
Team & Users 8 Profile, users, teams, working time
Messaging 4 Project + task messages
Allocations 3 CRUD
Ratecards 3 CRUD
Products 2 CRUD
Labels 1 By entity type

Quick Start

Prerequisites

  • Python 3.11+
  • uv (recommended) or pip
  • COR account with credentials

Install & Run

# Clone the repo
git clone https://github.com/YOUR_USER/cor-mcp-server.git
cd cor-mcp-server

# Install with uv (recommended)
uv sync

# Or with pip
pip install -e .

Configure

Copy .env.example to .env:

cp .env.example .env

Then choose one auth mode in .env:

Mode A — Email + Password (recommended for personal use):

COR_EMAIL=your.email@company.com
COR_PASSWORD=your_password
COR_API_URL=https://api.projectcor.com/v1

Mode B — API Key + Client Secret (for server-to-server):

COR_API_KEY=your_api_key
COR_CLIENT_SECRET=your_client_secret
COR_API_URL=https://api.projectcor.com/v1

The server auto-detects which mode you configured. Email/password takes priority when both are set.

Run

# stdio transport (default — for MCP agents)
uv run cor-mcp-server

# SSE/HTTP transport (for remote access)
uv run cor-mcp-server --transport sse --host 0.0.0.0 --port 8000

# Verbose logging
uv run cor-mcp-server --verbose

Authentication

Two modes supported. The server auto-detects which to use.

User Credentials (Mode A)

POST https://api.projectcor.com/v1/auth/login
Content-Type: application/x-www-form-urlencoded

email=user@corp.com&password=*****

Returns JWT access_token + refresh_token. Auto-refresh built in.

Client Credentials (Mode B)

POST https://api.projectcor.com/v1/oauth/token?grant_type=client_credentials
Authorization: Basic base64(COR_API_KEY:COR_CLIENT_SECRET)

Returns access_token (1-hour TTL). Cached and auto-refreshed with a 60-second safety margin.

Connecting from MCP Clients

Hermes Agent (recommended)

Add to ~/.hermes/config.yaml:

mcp_servers:
  cor:
    command: "uv"
    args: ["run", "--directory", "/path/to/cor-mcp-server", "python", "-m", "cor_mcp_server"]
    env:
      COR_EMAIL: "your.email@company.com"
      COR_PASSWORD: "your_password"
      COR_API_URL: "https://api.projectcor.com/v1"
    timeout: 180
    connect_timeout: 60

Then restart or run /reload-mcp.

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "cor": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/cor-mcp-server", "python", "-m", "cor_mcp_server"],
      "env": {
        "COR_EMAIL": "your.email@company.com",
        "COR_PASSWORD": "your_password",
        "COR_API_URL": "https://api.projectcor.com/v1"
      }
    }
  }
}

Cursor / Windsurf / Cline

In MCP Servers settings, add:

  • Name: COR
  • Type: command
  • Command: uv run --directory /path/to/cor-mcp-server python -m cor_mcp_server
  • Environment variables: COR_EMAIL, COR_PASSWORD, COR_API_URL

Claude Code

env COR_EMAIL=your@email.com COR_PASSWORD=your_pass COR_API_URL=https://api.projectcor.com/v1 \
  npx @anthropic/claude-code --mcp "uv run --directory /path/to/cor-mcp-server python -m cor_mcp_server"

Docker

Build with the included Dockerfile, pass env vars at runtime (supports either auth mode).

docker build -t cor-mcp-server .

# stdio mode
docker run -i --rm \
  -e COR_EMAIL=your@email.com \
  -e COR_PASSWORD=your_password \
  -e COR_API_URL=https://api.projectcor.com/v1 \
  cor-mcp-server

# SSE mode
docker run -i --rm \
  -p 8000:8000 \
  -e COR_EMAIL=your@email.com \
  -e COR_PASSWORD=your_password \
  -e COR_API_URL=https://api.projectcor.com/v1 \
  cor-mcp-server --transport sse

Available Tools

All tool names are prefixed with cor_ for clean discoverability.

Projects (14 tools)

Tool Description
cor_list_projects List projects with filters (clientId, status, health, dates)
cor_get_project Get project details
cor_create_project Create a new project
cor_update_project Update a project (partial)
cor_delete_project Delete a project
cor_get_project_collaborators Get project collaborators
cor_add_project_collaborator Add collaborator to project
cor_remove_project_collaborator Remove collaborator from project
cor_get_project_costs Get project costs/estimates
cor_add_project_cost Add cost to project
cor_get_project_labels Get project labels
cor_get_project_ratecard Get project ratecard
cor_get_project_templates Get available project templates
cor_get_project_profitability Get project profitability

Tasks (10 tools)

Tool Description
cor_search_tasks Search tasks with filters (projectId, clientId, status, text, dates, labels)
cor_get_my_pending_tasks Get my pending tasks
cor_get_task Get task details
cor_create_task Create a new task
cor_update_task Update a task
cor_delete_task Delete a task
cor_get_task_collaborators Get task collaborators
cor_sync_task_collaborators Sync (replace) task collaborators
cor_add_task_label Add label to task
cor_remove_task_label Remove label from task

Clients (6 tools)

Tool Description
cor_list_clients List clients
cor_get_client Get client details
cor_create_client Create a new client
cor_update_client Update a client
cor_delete_client Delete a client
cor_get_client_fees Get client fees

Contracts (5 tools)

Tool Description
cor_list_contracts List contracts
cor_get_contract Get contract details
cor_create_contract Create a new contract
cor_get_contract_positions Get contract positions
cor_create_contract_position Create position in contract

Time Tracking (5 tools)

Tool Description
cor_log_hours Log hours against a task
cor_search_time_entries Search time entries with filters
cor_get_hours_by_date Get time entries by date
cor_change_hours_status Change time entry status
cor_accept_suggested_hours Accept suggested hours

Team & Users (8 tools)

Tool Description
cor_get_my_profile Get current user profile
cor_list_users List users with filters
cor_get_user Get user details
cor_list_teams List teams
cor_create_team Create a new team
cor_add_team_users Add users to a team
cor_remove_team_users Remove users from a team
cor_get_working_time Get working time for users

Messaging (4 tools)

Tool Description
cor_get_task_messages Get messages on a task
cor_post_task_message Post a message on a task
cor_get_project_messages Get messages on a project
cor_post_project_message Post a message on a project

Ratecards (3 tools)

Tool Description
cor_list_ratecards List ratecards
cor_get_ratecard Get ratecard details
cor_create_ratecard Create a new ratecard

Allocations (3 tools)

Tool Description
cor_get_allocations_by_project Get allocations for a project
cor_save_allocation Create/update a resource allocation
cor_delete_allocation Delete a resource allocation

Products (2 tools)

Tool Description
cor_list_products List products
cor_create_product Create a new product

Labels (1 tool)

Tool Description
cor_get_labels Get labels (filter by entity type: project, task, user)

Architecture

┌──────────────────────┐     ┌───────────────────────────┐     ┌──────────────────────┐
│  MCP Client          │     │  COR MCP Server           │     │  COR REST API        │
│  (Hermes / Claude /  │◄───►│  (FastMCP + httpx)        │────►│  api.projectcor.com  │
│   Cursor / etc.)     │     │                            │     │  /v1                 │
└──────────────────────┘     ├───────────────────────────┤     └──────────────────────┘
                             │  Auth auto-detection       │
                             │  • Email/Password (prio)   │
                             │  • API Key + Secret        │
                             └───────────────────────────┘

Development

# Install dev deps
uv sync --dev

# Run tests (24 passing tests — no live API needed, uses mocking)
uv run pytest tests/ -v

# Test server startup
uv run cor-mcp-server --help

# Lint
uv run ruff check .

Adding New Tools

To add a new tool to the COR MCP Server:

  1. Create or edit a tool module in cor_mcp_server/tools/ (e.g., reports.py).

  2. Write an async function with the cor_ prefix pattern:

    from ..context import get_client
    
    async def cor_list_reports(page: int = 1, per_page: int = 20) -> str:
        """List reports.
    
        Args:
            page: Page number (default: 1)
            per_page: Results per page (default: 20)
        """
        client = get_client()
        data = await client.get("/reports", page=page, per_page=per_page)
        import json
        return json.dumps(data, indent=2, ensure_ascii=False, default=str)
    
  3. Register the tool in cor_mcp_server/server.py:

    from .tools import reports as _reports
    
    mcp.tool(
        name="cor_list_reports",
        description=_reports.cor_list_reports.__doc__,
    )(_reports.cor_list_reports)
    
  4. Add tests in tests/test_tools.py following the existing patterns (import assertion + tool count update).

  5. Update the tool count in tests/test_tools.py if needed.

The get_client() function provides a fully-authenticated CORClient instance with automatic Bearer token handling. All tools return JSON strings for consistent serialization.

Test Structure

Tests are in tests/test_tools.py and cover:

  • TestTokenStore — Thread-safe token caching, expiry, and refresh logic
  • TestCORClient — HTTP methods (GET, POST, PUT, DELETE), pagination, auth headers
  • TestToolFunctions — Import verification for all 61 tool functions across 11 modules
  • TestServerCreation — Server factory creates a valid FastMCP instance

All tests use mocking — no live COR credentials required to run them.

Project Structure

cor-mcp-server/
├── pyproject.toml              # Dependencies and metadata
├── README.md                   # This file
├── .env.example                # Env var template (both auth modes)
├── Dockerfile                  # Container deployment
├── skills/                     # Hermes Agent skills
│   └── cor-mcp-setup/
│       └── SKILL.md            # Skill: setup + usage guide
├── handoff-personal-hermes.md  # Self-contained instructions for personal Hermes
├── cor_mcp_server/
│   ├── __init__.py             # Package init
│   ├── __main__.py             # CLI entry point (argparse: --transport, --host, --port)
│   ├── server.py               # FastMCP server + tool registration (all 61 tools)
│   ├── auth.py                 # OAuth 2.0 dual-mode (user creds + client creds)
│   ├── client.py               # COR API HTTP client (httpx, async)
│   ├── context.py              # Shared client context / dependency injection
│   ├── models.py               # Pydantic models (14 entity types)
│   └── tools/
│       ├── __init__.py
│       ├── projects.py         # 14 tools
│       ├── tasks.py            # 10 tools
│       ├── time_tracking.py    # 5 tools
│       ├── clients.py          # 6 tools
│       ├── contracts.py        # 5 tools
│       ├── messaging.py        # 4 tools
│       ├── team.py             # 8 tools
│       ├── labels.py           # 1 tool
│       ├── ratecards.py        # 3 tools
│       ├── allocations.py      # 3 tools
│       └── products.py         # 2 tools
├── tests/
│   ├── __init__.py
│   └── test_tools.py           # 24 tests for auth, client, and tools
└── .venv/                      # Virtual environment (local)

License

MIT

Author

Alejandro Duran — Omnicom Media Group

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选