@mcp-abap-adt/calm-server
MCP server for SAP Cloud ALM, providing 54 tools across 9 services to manage features, tasks, test cases, documents, projects, and more via natural language.
README
@mcp-abap-adt/calm-server
MCP server for SAP Cloud ALM, built on
@mcp-abap-adt/calm-client.
Ships 54 MCP tools covering all 9 Cloud ALM services — full CRUD
where the service supports it, plus a wide read surface — with rich
JSON Schema descriptions that let an LLM plan multi-step workflows.
This package is dual-purpose:
- Runnable stdio server —
npx @mcp-abap-adt/calm-server. Plug directly into Claude Desktop, Claude Code, or any MCP-compatible host. - Reusable library of tool primitives — import individual tools
(
FEATURES_GROUP,TASKS_GROUP, …) and embed them in a larger composed MCP server without reimplementing anything.
Status
0.2.0 — 54 tools, 232 tests (224 unit + integration, 7 env-gated
skips, 1 todo), full build green. Integration suite runs live against
the SAP sandbox (api.sap.com) or any OAuth2 Cloud ALM tenant with a
single .env switch; gates skip cleanly when no backend is wired.
Installation
As a standalone MCP server
npm install -g @mcp-abap-adt/calm-server
# or per-project:
npm install @mcp-abap-adt/calm-server
As a library (compose into your own MCP server)
npm install @mcp-abap-adt/calm-server
# peers:
npm install @mcp-abap-adt/calm-client @mcp-abap-adt/interfaces @modelcontextprotocol/sdk
Standalone: running the server
1. Configure credentials
Copy the template and fill in:
cp .env.example .env
OAuth2 mode (real tenant) — paste values from an XSUAA service key:
CALM_MODE=oauth2
CALM_BASE_URL=https://<tenant>.<region>.alm.cloud.sap
CALM_UAA_URL=https://<tenant>.authentication.<region>.hana.ondemand.com
CALM_UAA_CLIENT_ID=sb-…!b…|calm!b…
CALM_UAA_CLIENT_SECRET=…
Sandbox mode (SAP API Business Hub):
CALM_MODE=sandbox
CALM_API_KEY=<your-key-from-api.sap.com>
# CALM_BASE_URL defaults to https://sandbox.api.sap.com/SAPCALM
2. Launch
# Global install:
calm-mcp
# Or via npx without install:
npx @mcp-abap-adt/calm-server
# Or from a clone:
npm run build && node dist/bin/stdio.js
The server speaks MCP over stdio. Misconfiguration is reported to
stderr with a non-zero exit code.
Authentication setup
calm-mcp supports two OAuth2 flows for live tenants, selectable via
CALM_AUTH_FLOW:
| Flow | Use case | Browser? | Refresh token? |
|---|---|---|---|
client_credentials (default) |
technical service-binding (sb-* client) |
no | no |
authorization_code |
end-user dev workflow, full user scope | once | yes |
Option A — quick CC setup (technical user)
Plain .env with inline CALM_UAA_URL / CALM_UAA_CLIENT_ID /
CALM_UAA_CLIENT_SECRET works as before — no extra steps. The broker uses
an in-memory session shim for these inline creds.
Option B — broker-backed setup (CC or AC)
Use the bundled mcp-auth
CLI to convert a BTP service key into a token-bearing .env:
# CC (no browser)
npx mcp-auth --service-key ./sk.json --output ./DEFAULT.env \
--type xsuaa --credential
# AC (browser pops once; refresh_token persists)
npx mcp-auth --service-key ./sk.json --output ./DEFAULT.env \
--type xsuaa --browser auto
Then in your .env:
CALM_MODE=oauth2
CALM_BASE_URL=https://<tenant>.<region>.alm.cloud.sap
CALM_AUTH_FLOW=authorization_code # or client_credentials
CALM_DESTINATION=DEFAULT
The server's runtime auth pipeline is @mcp-abap-adt/auth-broker.
Note:
buildCalmClientis async since v0.4.0 (was sync in v0.3.x). Library consumers mustawaitit.
3. Wire into Claude Desktop
Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json
on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows):
{
"mcpServers": {
"calm": {
"command": "npx",
"args": ["-y", "@mcp-abap-adt/calm-server"],
"env": {
"CALM_MODE": "sandbox",
"CALM_API_KEY": "<your-sandbox-key>"
}
}
}
}
Restart Claude Desktop; the 54 calm_* tools become available to the
model.
Library: composing into another MCP server
Useful when you want to expose Cloud ALM tools alongside ADT tools, Reports tools, or your own domain tools in a single MCP process.
import { CalmClient, CalmConnection } from '@mcp-abap-adt/calm-client';
import { McpServer } from '@modelcontextprotocol/sdk/server/mcp.js';
import {
ALL_GROUPS,
BaseCalmMcpServer,
CalmToolRegistry,
FEATURES_GROUP,
TASKS_GROUP,
} from '@mcp-abap-adt/calm-server';
// Option A — BaseCalmMcpServer with a curated subset
const calm = new CalmClient(
new CalmConnection({ baseUrl, apiKey }),
);
const server = new BaseCalmMcpServer({
name: 'my-mcp',
version: '1.0.0',
calm,
groups: [FEATURES_GROUP, TASKS_GROUP], // only these land as tools
});
// Option B — embed into your existing McpServer
const existing = new McpServer({ name: 'combined', version: '1.0.0' });
const registry = new CalmToolRegistry([...ALL_GROUPS]);
registry.registerAll(existing, () => ({ calm }));
// Now `existing` serves Cloud ALM tools + whatever else you registered.
Subpath exports:
import { ALL_GROUPS } from '@mcp-abap-adt/calm-server/tools';
import { CalmToolRegistry } from '@mcp-abap-adt/calm-server/registry';
Tool surface (54 tools across 9 services)
| Service | Tools |
|---|---|
| Features (11) | list, get, get_by_display_id, create, update, delete, create_external_reference, delete_external_reference, list_external_references, list_statuses, list_priorities |
| Tasks (10) | list, get, create, update, delete, list_comments, create_comment, list_references, list_deliverables, list_workstreams |
| TestCases (9) | list, get, create, update, delete, create_activity, create_action, list_activities, list_actions |
| Documents (7) | list, get, create, update, delete, list_statuses, list_types |
| Projects (7) | list, get, create, list_programs, get_program, list_team_members, list_timeboxes |
| Hierarchy (5) | list, get_with_children, create_node, update_node, delete_node |
| Analytics (2, read-only) | query (17 endpoints), list_providers (static catalog) |
| Logs (2, domain-specific REST) | get (provider + serviceId + time window), post (inbound log records) |
| ProcessMonitoring (1, read-only) | list_processes |
Every MCP tool:
- Has a full JSON Schema with descriptions — the LLM reads these to plan.
- Wraps arguments into OData
$filter/$select/$top/$skipinternally — the LLM never sees raw OData syntax. - Returns compact records by default (
limit=20,fieldsdefault ≈ 6 columns per entity); callers opt into more viafields,limit,withCount,offset. - Maps
CalmApiError→ a typed MCP error the LLM can branch on (NOT_FOUND,NETWORK,ODATA_ERRORwithserviceCode, …).
See src/tools/<service>/*.ts for per-tool argument schemas.
Destructive tools (write operations)
Every Cloud ALM service that supports writes is now exposed:
- Features:
create,update,delete, plus external-referencecreate/delete - Tasks:
create,update,delete,create_comment - TestCases:
create,update,delete, plus nestedcreate_activity/create_action - Documents:
create,update,delete - Hierarchy:
create_node,update_node,delete_node - Projects:
create - Logs:
post(inbound OpenTelemetry-style record ingestion)
The shared SAP sandbox at api.sap.com is read-friendly only — mutation
integration tests are opt-in (see Live-tenant integration below).
Debug logging
CALM_LOG_LEVEL=debug # error | warn | info | debug
DEBUG_CALM_CONNECTORS=true # CalmConnection retries, 401 refresh, URLs
DEBUG_CALM_LIBS=true # resource-client internals
DEBUG_CALM_TESTS=true # test execution progress
Logging goes to stderr (stdout is reserved for the MCP protocol stream).
Development
git clone git@github.com:fr0ster/mcp-calm-server.git
cd mcp-calm-server
npm install
npm run test # 224 unit + integration tests
npm run build # emits dist/, includes executable bin
npm run lint:check # biome
Live-tenant integration
src/__tests__/integration/ runs against a real backend when env is
present, and skips cleanly when it isn't (so npm test without secrets
is always green). Five gates, drop them into .env:
| Gate | Env trigger | What it unlocks |
|---|---|---|
describeSandbox |
CALM_MODE=sandbox + CALM_API_KEY |
The api.sap.com sandbox |
describeOAuth2 |
CALM_MODE=oauth2 + CALM_BASE_URL + 3× UAA env |
Live tenant (incl. endpoints absent from the sandbox catalog, e.g. Business Processes) |
describeWhenLive |
either of the above | Read-side tests that work in either mode |
describeWithProject |
live backend + CALM_PROJECT_ID |
Project-scoped chains (features list→get, tasks list→get→comments, …) |
describeMutating |
live + CALM_PROJECT_ID + CALM_ALLOW_MUTATIONS=1 |
Write tests (every mutation finalises via try/finally { delete }) |
A quick smoke script lives at scripts/smoke-sandbox.mjs — spawns the
stdio bin, lists tools, calls a handful of read endpoints, and exits
1 on any non-skip failure.
License
MIT — see LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.