Configure

Essential Environment Variables (.env file)

# SAP AI Core
SAP_CLIENT_ID=your_client_id
SAP_CLIENT_SECRET=your_client_secret
SAP_AUTH_URL=https://your-auth-url
SAP_AI_API_URL=https://your-ai-api-url
SAP_IDENTITY_ZONE=your_identity_zone
 
# Database
DB_CONFIG_DISCOVERY_URL_BASE=https://your-config-server
 
# Application Settings
DEFAULT_MODEL=gpt-4o
DEFAULT_TEMPERATURE=0.3
DEFAULT_MAX_TOKENS=1500
 
# Optional Integrations
MICROSOFT_CLIENT_ID=your_ms_client_id
SHAREPOINT_SITE_URL=https://your-sharepoint-site

Configuration File Reference

File	Purpose
.env	Environment variables and secrets
config.py	Application configuration classes
manifest.yaml	Deployment configuration (CF push settings)
config/	Environment-specific settings and prompts

Key Configuration Classes

Class	Purpose
SAPAICoreConfig	SAP integration settings and model defaults
ConfigCache	Performance-optimized configuration caching
ConfigLoader	Dynamic configuration loading per request

Multi-Tenant Configuration

Each tenant environment (DEV, SBX, PROD) can have its own:

• HANA database connection
• FAISS vector store
• Prompt templates and model settings
• SharePoint and integration credentials Tenant resolution is automatic — driven by the JWT token presented at runtime.

---

Tenant Configuration Files (Config Server)

All per-environment configuration is stored on the CherryWork Config Server and loaded dynamically at runtime. The folder structure on the config server mirrors the tenant layout shown below.

Folder Structure:

cherry-bot/
├── apa_dev/
├── itm/
├── scp_dev/
│   ├── db_configs.json
│   └── prompt_template.txt
├── soa_dev/
└── soa_qa/
application_properties.json
applications.json

---

applications.json — Tenant Registry

This is the master registry that maps each tenant key to its config URLs (DB config, MCP server, queries, etc.). The app reads this file at startup to discover all tenant environments.

{
    "itm": [
        "https://cherryworkproducts-config-server.../cherry-bot/itm/queries.txt",
        "https://cw-mcp-server.cfapps.eu10-004.hana.ondemand.com/mcp/nwf",
        "...similar_queries.yaml"
    ],
    "scp_dev": [
        "https://cherryworkproducts-config-server.../cherry-bot/scp_dev/db_configs.json",
        "https://cw-mcp-server.cfapps.eu10-004.hana.ondemand.com/mcp/scp",
        ""
    ],
    "scp_qa": [
        "https://cherryworkproducts-config-server.../cherry-bot_qa/scp_qa/db_configs.json",
        "https://cw-mcp-server-qa.cfapps.eu10-004.hana.ondemand.com/mcp/scp",
        ""
    ],
    "soa_qa": [
        "https://cherryworkproducts-config-server.../cherry-bot_qa/soa_qa/db_configs.json",
        "https://cw-mcp-server.cfapps.eu10-004.hana.ondemand.com/mcp/soa",
        "...queries.txt",
        "...similar_queries.yaml"
    ],
    "CHERRY_BOT_APP_KEY": "itm"
}

Array index meaning per tenant entry:

Index	Content
[0]	db_configs.json URL — HANA database connection for that tenant
[1]	MCP server URL — external tool/service integration endpoint
[2]	queries.txt URL — sample/similar queries for that tenant (if applicable)
[3]	similar_queries.yaml URL — semantic query matching config (if applicable)

CHERRY_BOT_APP_KEY sets the default active tenant when no tenant is specified in the JWT. Currently set to "itm".

To add a new tenant environment, add a new key to this file with its corresponding config URLs and redeploy (or trigger a config refresh).

---

db_configs.json — HANA Database Connection (Per Tenant)

Each tenant folder contains its own db_configs.json with isolated HANA credentials and table definitions.

{
    "database": "hana",
    "hana": {
        "hana_address": "<tenant-hana-host>.hanacloud.ondemand.com",
        "hana_port": 443,
        "hana_encrypt": "true",
        "hana_user": "CW-SCP-DEV",
        "hana_sslValidateCertificate": "True",
        "hana_password": "<password>"
    },
    "TABLES_TO_EXTRACT": [
        "CW_SCP_PO_HEADER_TXN",
        "CW_SCP_PO_LINE_ITEM_TXN"
    ]
}

Field reference:

Field	Description
database	Database type — always "hana"
hana_address	HANA Cloud instance hostname
hana_port	Always 443 for HANA Cloud
hana_encrypt	Must be "true" for cloud connections
hana_user	DB user for this tenant environment
hana_sslValidateCertificate	SSL validation — set "True" in production
hana_password	DB user password (store securely; never commit to Git)
TABLES_TO_EXTRACT	List of HANA tables the SQL Generator is allowed to query for this tenant

Each tenant environment (DEV, QA, PROD) has its own db_configs.json with separate credentials and table lists.

---

prompt_template.txt — Environment Prompt (Per Tenant)

Each tenant folder optionally contains a prompt_template.txt which defines the system prompt and behavior instructions for the LLM in that environment. This allows per-tenant customization of tone, domain focus, and response style without any code changes.

To update the prompt for a tenant: edit the prompt_template.txt in the corresponding folder on the Config Server and trigger a config reload (or restart the app).

MCP Server Configuration

Environment Variables

All configuration is provided via environment variables — no code changes are needed across environments.

Variable	What It Controls	Example
FAISS_INDEX_PATH	Location of the FAISS vector index file	faiss_index/soa_dev/index.faiss
DB_CONN	Database connection string	jdbc:sap://...
API Keys	Authentication for model and service backends	Set per environment
Model Endpoints	Which LLM backend to call	AI Core service URL

Environment-Based Service Routing (config.yaml)

Controls which backend URLs are used per environment. Change current_environment to switch routing without touching any tool definitions.

service_name: "soa"
display_name: "Sales Order Management"
environments:
  dev:
    destinations:
      soa:
        base_url: "https://dev-service-url"
  qa:
    destinations:
      soa:
        base_url: "https://qa-service-url"
current_environment: "dev"   # Change this to switch environments

Defining Tools (tools.yaml)

Each tool is declared with its name, description, HTTP method, endpoint, and parameters:

version: "1.0"
tools:
  - name: get_top_item_quantity
    description: Fetch top-selling items within a date range
    destination: soa
    method: GET
    endpoint: /api/dashboard/getTopItemQuantity
    parameters:
      - name: startDate
        type: string
        required: true
        description: Start date (MMDDYYYY)
        location: query
      - name: endDate
        type: string
        required: true
        description: End date (MMDDYYYY)
        location: query
      - name: limit
        type: integer
        required: false
        default: 10
        location: query
    response:
      type: json

Configuration File Reference

File	Purpose
config.yaml	Environment-based service routing
tools.yaml	Tool definitions (name, endpoint, parameters)
manifest.yaml	Service and tool metadata registry
db_config.py	Database configuration

---