procare-mcp

procare-mcp

First MCP server for the Procare Connect API, enabling querying child care center data (attendance, enrollments, family balances, etc.) via natural language from Claude, Cursor, or any MCP client.

Category
访问服务器

README

procare-mcp

First MCP server for the Procare Connect API — talk to your child care center data from Claude, Cursor, or any MCP client.

Procare is the leading child care management platform used by 30,000+ daycare centers, preschools, and before/after-school programs in the US. The public Connect API is documented at https://api-docs.procareconnect.com/ and powers the Procare Online + SchoolCare Works products. This server wraps that API in the Model Context Protocol so you can ask your AI agent questions like "which children at school 1234 were absent on Monday?" or "what's the balance on family 555?" — without hand-rolling a REST client.

What you can do with it

You:   "Pull the attendance log for school 1234 for last week and tell me
        which kids had 3+ absences."
Claude: *calls list_schools, then list_attendance with school_id+date_from
        +date_to, summarises the result*

You:   "Show me the enrollment roster for the summer camp at our
        Westside location."
Claude: *calls list_programs then list_registrations with school_id +
        program_id filters*

You:   "What's the outstanding balance for family 555?"
Claude: *calls get_family, returns the balance_cents field with a
        human-readable summary*

Install

pip install -e .

Configure

Procare uses OAuth 2 client_credentials against your tenant subdomain, not a shared host. Find your subdomain by looking at the URL of the Procare dashboard you log into — it'll be https://<your-subdomain>.procareconnect.com. Your Procare account admin can mint a client_id / client_secret pair with API access enabled.

export PROCARE_BASE_URL="https://<your-subdomain>.procareconnect.com"
export PROCARE_CLIENT_ID="..."
export PROCARE_CLIENT_SECRET="..."

Who uses this?

  1. Child care center owners and operators who want to query their data via Claude / Cursor without building a custom integration.
  2. Multi-location operators (5+ centers) who need to roll up attendance, payments, and enrollment across the enterprise.
  3. API Partners building tools on top of Procare.

If you don't have API credentials yet, contact your Procare account manager or open a ticket with Procare support.

Use with Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent path on Windows / Linux:

{
  "mcpServers": {
    "procare-mcp": {
      "command": "procare_mcp",
      "env": {
        "PROCARE_BASE_URL": "https://your-subdomain.procareconnect.com",
        "PROCARE_CLIENT_ID": "your-client-id",
        "PROCARE_CLIENT_SECRET": "your-client-secret"
      }
    }
  }
}

Use with Claude Code

claude mcp add procare-mcp -- procare_mcp \
  --env PROCARE_BASE_URL=https://your-subdomain.procareconnect.com \
  --env PROCARE_CLIENT_ID=your-client-id \
  --env PROCARE_CLIENT_SECRET=your-client-secret

Tools

Tool Type What it does
health_check Diagnostic Verifies credentials by listing schools (counts them)
list_schools Read All schools/locations in the enterprise account
get_school Read Single school by id
list_children Read Children, filterable by school_id / paginated via offset+limit
get_child Read Single child by id (allergies, schedule, classroom)
list_families Read Families (households), filterable by school_id / paginated
get_family Read Single family by id (guardians, children, contacts, balance)
list_attendance Read Attendance records, filterable by school / child / date range
list_programs Read Programs (camps, preschool, after-school), filterable by school / active
get_program Read Single program by id (description, schedule, capacity, fee)
list_registrations Read Enrollments, filterable by program / child / school
list_payments Read Payments, filterable by family / school / date range
list_staff Read Staff members, filterable by school / active flag
get_staff_member Read Single staff member by id
list_classrooms Read Classrooms, filterable by school
get_classroom Read Single classroom by id (capacity, age range, lead teacher)

The read-only cut covers every endpoint exposed by the Connect API that the v0.1 surface needs. Write tools (check-in, check-out, create registration) will land in a follow-up once the upstream write contract is stable across Procare Online and SchoolCare Works.

Engineering

This server follows the industry-leading patterns baked into mcp-vertical-template:

  • Shared httpx.AsyncClient with connection pooling and transport-level retries
  • OAuth 2 client_credentials with proactive refresh (5-min buffer) and an asyncio.Lock to serialize concurrent refreshes
  • Typed exception hierarchy (ProcareAuthError, ProcareNotFoundError, ProcareRateLimitError, ProcareAPIError, ProcareConnectionError) with structured fields (http_status, error_code, request_id, retry_after)
  • Dispatch table for HTTP status → exception (no if/elif chains)
  • Application-level retry with exponential backoff + full jitter on 429/5xx, honoring Retry-After
  • Single 401 retry that force-refreshes the OAuth token (handles server-side revocation)
  • JSONL audit logging per tool call (stderr default, PROCARE_AUDIT_LOG env var override) with secret redaction
  • Bare raise inside MCP tool handlers — isError=true on the wire, per the Blackwell MCP security audit baseline
  • respx mocks + hypothesis property tests (no live API in CI)
  • ruff full rule set + mypy --strict
  • CI matrix on Python 3.10-3.13 × Ubuntu/macOS/Windows

Development

python -m venv .venv && source .venv/bin/activate
pip install -e ".[dev]"
pytest -v
procare_mcp

License

MIT.

See also

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