Crescender MCP Server
Crescender MCP 服务器
用于 Crescender 公共只读 API 的模型上下文协议 (MCP) 服务器。它将 AI 客户端(Claude Desktop、Cursor、自定义代理)连接到学校的资产、借用、成员和资产通信线程——所有操作均为只读、通过范围令牌认证,且每个令牌绑定到一所学校。
状态:0.0.x — 发布前版本。 此包尚未发布到 npm 或推送到 GHCR。发布流水线已存在,但仅限手动触发;首个版本将在源代码经过本地压力测试后发布。目前,请从克隆版本或使用
pnpm link运行(参见下方的 本地开发)。
它为 AI 提供了什么
六个只读工具,每个工具均由 /api/v1/* 端点支持:
MCP 工具 | 返回值 |
| 令牌绑定的唯一学校(单项数组) |
| 按 ID 获取完整资产详情 — 型号、序列号、位置、角色、当前借用情况 |
| 最多 50 个匹配 |
| 单个资产的借用历史(当前 + 过去) |
| 成员目录 — ID、角色、状态、external_id(无个人隐私信息 PII) |
| 带有状态过滤器的资产通信线程(只读 — 供应商令牌颁发不通过 MCP 公开) |
仅限读取。v1 版本中没有 update_* / create_* 工具。
Related MCP server: OfficeRnD MCP Server
配置
两个环境变量,其中一个是必需的:
变量 | 默认值 | 说明 |
| — | 必需。 从 |
|
| 仅在指向非生产环境部署时覆盖。 |
|
|
|
|
| 在容器部署中设置为 |
|
| 每个请求的上游超时时间。 |
连接 Claude Desktop(发布后)
一旦第一个版本发布到 npm:
// ~/Library/Application Support/Claude/claude_desktop_config.json (macOS)
// %APPDATA%\Claude\claude_desktop_config.json (Windows)
{
"mcpServers": {
"crescender": {
"command": "npx",
"args": ["-y", "@crescender/mcp-server"],
"env": {
"CRESCENDER_API_TOKEN": "crsc_..."
}
}
}
}然后重启 Claude Desktop。这六个工具应该会出现在 MCP 选择器中。
连接 Cursor(发布后)
Cursor 从工作区中的 .cursor/mcp.json 或全局设置中读取 MCP 配置。其结构与 Claude Desktop 的 mcpServers 块相同。
本地开发
# Install
git clone https://github.com/lincalinca/crescender-mcp-server.git
cd crescender-mcp-server
pnpm install
# Build
pnpm build
# Smoke test (talks to stdin/stdout — Claude Desktop / Cursor will spawn it
# the same way)
CRESCENDER_API_TOKEN=crsc_... node dist/index.js要在不发布的情况下将本地克隆版本连接到 Claude Desktop:
{
"mcpServers": {
"crescender-local": {
"command": "node",
"args": ["/absolute/path/to/crescender-mcp-server/dist/index.js"],
"env": {
"CRESCENDER_API_TOKEN": "crsc_..."
}
}
}
}容器
# Build (the published image is built by .github/workflows/publish-image.yml)
docker build -t crescender-mcp-server:dev .
# Run with the in-container /health endpoint enabled
docker run --rm -i \
-e CRESCENDER_API_TOKEN=crsc_... \
-e CRESCENDER_MCP_HEALTH=1 \
-p 3030:3030 \
crescender-mcp-server:dev发布后,镜像将可在 ghcr.io/lincalinca/crescender-mcp-server 获取。
无状态契约
无内存令牌缓存:验证在每次调用时都会访问上游 API(撤销是即时的)。
无文件系统状态:仅限 stdout/stderr。
配置由环境变量驱动,在启动时解析一次并冻结。
/health(启用时)返回{ ok, api_reachable, version }并探测上游/api/v1/health以验证可达性 — 而不仅仅是“此进程存活”。优雅关闭:SIGTERM / SIGINT 触发 30 秒的排空期,然后强制退出。
这就是使容器无需任何 k8s 特定代码即可实现 k8s 就绪的原因。当最终做出编排决策(ASK / AKS / GKE / Fly / Railway 等)时,此处无需更改任何内容。
许可证
MIT — 参见 LICENSE。
相关
@crescender/crescender-core— 平台本身,提供此 MCP 封装的/api/v1端点。OpenAPI 规范:
https://app.crescender.com.au/api/v1/openapi.json
Available Tools
6 toolsget_assetA
Get full detail for a single asset (instrument, equipment, sub-item) by id. Includes model, serial, location, role, and the current loan if any.
Inputs: asset_id (required).
get_loans_for_assetA
Get the loan history for a single asset. Currently returns whatever loans field is embedded on the asset detail; in a future version may paginate independently.
Inputs: asset_id (required).
list_asset_threadsA
List asset-comms threads (vendor / repair / service conversations attached to assets). Read-only — vendor-token issuance is NOT exposed via this MCP.
Inputs: status, cursor.
list_membersA
List the school's members. Returns id, role, status, external_id only — no PII (email/phone are intentionally excluded from the public API).
Inputs: role, status, cursor.
list_schoolsA
List schools accessible to the calling token. Each token is bound to exactly one school, so this returns a single-item array. Useful for the AI to confirm which school it's looking at before making other calls.
Inputs: no named parameters.
search_assetsA
Search the school's assets by free-text query and/or filters. Returns up to 50 matches per page. Use the cursor field from the response to fetch the next page.
Inputs: q, item_class, category, status, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 6 tools
Each tool targets a distinct resource or action: get_asset and search_assets are clearly different (detail vs. search), and loan history, threads, members, and schools are all separate concerns. No overlapping purposes.
All tools consistently follow the 'verb_noun' pattern in lowercase snake_case (e.g., get_asset, list_asset_threads, search_assets). The naming is uniform and predictable.
6 tools is well-scoped for an asset management server. Each tool covers a core area without being overwhelming or insufficient.
The tool surface is heavily read-only: assets have get and search but no create/update/delete; loans only show history; threads only list; members and schools only list. Missing basic write operations for the domain.
Maintenance
Related MCP Connectors
Read-only MCP server for ClassQuill, a tutoring-business-management platform.
Read-only MCP server exposing a user ORANO library to their own AI agent.
Read-only MCP server for The Quiet Protocol's engines, benchmarks, proof, and business data.
Read-only MCP server for the OPERANT AI operating-agent calibration benchmark.
Related MCP Servers
- AlicenseAqualityDmaintenanceA read-only MCP server that provides access to Charles Schwab account data and market information, including portfolio positions, real-time quotes, options chains, price history, and account balances through AI assistants.9MIT
- AlicenseAqualityDmaintenanceA read-only MCP server that connects AI assistants to the OfficeRnD coworking and flex-space management platform. It enables natural language queries for community members, space bookings, billing records, and office resources.51MIT
- AlicenseAqualityDmaintenanceRead-only MCP server for Litmos LMS that exposes user profiles, course results, and training completion data to AI clients.8MIT

milele-prime-mcpofficial
FlicenseNot gradedqualityDmaintenanceRead-only MCP server that lets a verified client connect their own AI to their trading account to read account data without order execution.-