MCP Server logo

MCP Server

by cap-js

Searches local CAP documentation and compiled CDS models to support AI-assisted application development.

Install MCP Server

Pick your client, then add it in one click or paste the config.

STDIONode.js
~/Library/Application Support/Claude/claude_desktop_config.json
claude_desktop_config.json
{
  "mcpServers": {
    "mcp-server-107": {
      "command": "npx",
      "args": [
        "-y",
        "@cap-js/mcp-server"
      ]
    }
  }
}

Settings → Developer → Edit Config, merge this into mcpServers, then fully quit and reopen Claude Desktop.

Install not yet sandbox-tested— we haven't run this command through our automated sandbox yet. Not a red flag; we're working through the catalog.

Overview

cap-js/mcp-server provides MCP tools for SAP Cloud Application Programming Model (CAP) development. It searches locally stored CAP documentation with vector embeddings and performs fuzzy searches across definitions in a compiled CDS model, including relationships, annotations, and HTTP endpoints. Run it through an MCP client with npx or invoke its tools from the `cds-mcp` command-line interface. Reach for it when an AI coding assistant needs project-specific CDS information or CAP guidance before changing models or using CAP APIs.

Use cases

  • Search CAP documentation before changing a CDS model
  • Find entities and services in a compiled project model
  • Locate CAP HTTP endpoints and model relationships
  • Query CAP guidance from an AI coding client
  • Run documentation and model searches from the command line

Key features

  • Semantic search over local CAP documentation
  • Fuzzy search across compiled CDS definitions
  • Runtime-aware documentation filtering
  • Optional local cross-encoder reranking
  • MCP and command-line invocation

Tools2

What MCP Server lets your agent do. Expand a tool for its parameters.

Self-reported~70 tokens of context

Security & health

What it can reach and how reliably it runs — from our own checks, not the maintainer’s claims.

ActiveLast checked Today

Recent health check succeeded.

8/8 checks healthy over the last 40d

Supply-chain signal

No high-severity advisories surfaced by our automated scan.

Critical 0High 0Medium 0Low 0

Scanned 15d ago via OSV.dev · @cap-js/mcp-server (npm)

Adoption & maintenance

Factual signals from GitHub, npm, and our automated checks — not a rating.

GitHub stars117Stargazers on the source repository.
npm downloads211kPackage downloads in the last 30 days.
Last commitTodayMost recent push to the default branch.
Tools exposed2Callable tools this server registers over MCP.
Directory activity4 viewsConfig copies, upvotes, and views on AllMCPs.

How MCP Server works

What cap-js/mcp-server does

cap-js/mcp-server connects AI coding clients to SAP Cloud Application Programming Model projects. Its two MCP tools address different kinds of questions: search_model finds definitions in the project's compiled Core Schema Notation (CDS) model, while search_docs retrieves relevant passages from locally stored CAP documentation.

The model search is useful for locating entities, fields, services, relationships, annotations, and HTTP endpoints. The documentation search supports questions about CAP development, such as changing a Node.js query or using the cds command-line interface. The project recommends using these searches when working with CDS definitions and CAP APIs.

How it works

search_docs converts a query into an embedding vector and compares it with preprocessed documentation chunks stored as embeddings. Results are selected by semantic similarity, so the query does not need to use the exact wording found in the documentation. By default, documentation results are filtered to the detected project runtime, either Node.js or Java, together with runtime-independent content. The CDS_MCP_GENERIC_EMBEDDINGS setting disables that filter.

search_model performs fuzzy matching against names in the compiled CDS model. CDS combines the project's .cds files into a unified representation, allowing the tool to search partial names such as book for a definition named Books.

Setup and configuration

An MCP client can start the server with:

Terminal
npx -y @cap-js/mcp-server

The repository includes configuration examples for Cline in VS Code, VS Code's mcp.json, Claude Code, OpenAI Codex, and opencode. It also provides a command-line installation path:

Terminal
npm i -g @cap-js/mcp-server

A global installation supplies the cds-mcp command. Global users must update the package manually because npx uses the installed version instead of downloading the newest package.

Configuration options include CDS_MCP_OFFLINE for using cached embeddings without checking for updates, CDS_MCP_MODEL for selecting the embedding model, and timeout variables for manifest and bundle requests. Search result limits can be controlled with CDS_MCP_MAX_RESULTS. Optional local reranking uses RERANK_ENABLED together with the CDS_MCP_RERANK_* settings.

The tools can also be called directly:

sh
cds-mcp search_docs "how to add columns to a select statement in CAP Node.js" 1
cds-mcp search_model . Books entity

Tools and capabilities

  • search_docs: Semantically searches locally prepared CAP documentation using vector embeddings.
  • search_model: Fuzzy-searches names in the compiled CDS model.
  • Searches can cover CDS entities, fields, services, annotations, relationships, and HTTP endpoints through the compiled model.
  • Documentation retrieval can use an optional local cross-encoder reranker when enabled.

Limitations and notes

The tools are intended for AI models and are not presented as a stable API. Documentation embeddings and optional reranking models are handled locally, with settings available for offline operation and model selection. The recommended agent rules require searching for CDS definitions before reading .cds files and searching CAP documentation before creating or modifying CDS models or using CAP APIs or the cds CLI.

Reviews

No reviews yet — be the first to share how it worked for you.

Frequently Asked Questions about MCP Server

cap-js/mcp-server is an MCP server that connects AI coding clients to SAP Cloud Application Programming Model projects. Its main tools, `search_docs` and `search_model`, search local CAP documentation and compiled CDS model definitions, including entities, services, relationships, annotations, and HTTP endpoints.

Run `npx -y @cap-js/mcp-server` from the MCP client configuration, or install it globally with `npm i -g @cap-js/mcp-server`. A global installation provides the `cds-mcp` command, but the package must be updated manually.

It provides `search_docs` and `search_model`. `search_docs` uses vector embeddings to find relevant local CAP documentation, while `search_model` performs fuzzy searches against definitions in the compiled CDS model.

Install the package globally, then invoke `cds-mcp search_docs` for documentation queries or `cds-mcp search_model` for CDS definitions. For example, `cds-mcp search_model . Books entity` searches the model in the current project.

Cursor is not among the clients configured in the documented examples. The documented setup includes Cline, VS Code, Claude Code, OpenAI Codex, and opencode, using the command `npx -y @cap-js/mcp-server`.

Related MCP servers

Add a badge to your READMEShow it’s listed — live SVG, Markdown or HTML.

AllMCPs Directory Badge

Full Badge Customizer

Showcase your server listing on GitHub or your project documentation. Embed this dynamic SVG badge to highlight official listing status and live engagement.

Badge Style:
Live Dynamic SVG PreviewMCP Server AllMCPs Directory Badge
Markdown (GitHub README)
[![AllMCPs](https://allmcps.com/api/badge/mcp-server-107?style=directory)](https://allmcps.com/mcp/mcp-server-107)
HTML Embed
<a href="https://allmcps.com/mcp/mcp-server-107"><img src="https://allmcps.com/api/badge/mcp-server-107?style=directory" alt="MCP Server on AllMCPs" /></a>