OVERVIEW

Sage X3 (formerly Sage Enterprise Management) is a mid-market ERP powering manufacturing, distribution, and services companies. Modern instances expose a GraphQL API via the Syracuse web server using the Yoga GraphQL engine.

This skill gives AI agents the knowledge to query, report on, and integrate with Sage X3 — without hallucinating endpoints or field names.

ARCHITECTURE

The recommended integration pattern uses a Cloudflare Worker as a proxy — this avoids CORS, secures credentials at the edge, and enables AI analysis via Workers AI.

Endpoint Pattern

https://<host>:<port>/<folder>/api

Authentication

Authorization: Basic base64(USERNAME:PASSWORD)

CAPABILITIES

What your agent can do with this skill installed:

• Query master data — customers, suppliers, products, stock sites
• Pull transactions — sales orders, purchase orders, invoices, deliveries
• Check inventory — stock levels, reorder alerts, warehouse positions
• Generate reports — revenue by customer, aged debtors, sales trends
• Introspect the schema — discover what's available on any Sage X3 instance
• Build dashboards — structure data for Chart.js, tables, and KPI cards
• Natural language queries — translate questions into GraphQL
• Deploy to edge — Cloudflare Workers proxy with AI analysis

SCHEMA MAP

Sage X3 organizes its GraphQL API by business domain. Each domain contains entities that follow the Relay pagination pattern.

QUERY PATTERNS

Relay Pagination

All list queries use edges / node with cursor-based pagination:

query {
  x3MasterData {
    customer {
      query(first: 10, after: "cursor") {
        edges {
          node {
            code {
              code
              companyName1
              country { code countryName }
              currency { code }
            }
          }
        }
        pageInfo {
          hasNextPage
          endCursor
        }
      }
    }
  }
}

Customer Master Data

query Customers {
  x3MasterData {
    customer {
      query(first: 50) {
        edges {
          node {
            code {
              code
              companyName1
              country { code countryName }
              currency { code localizedDescription }
              isCustomer
              isSupplier
              creditLimit
              paymentTerms { code localizedDescription }
            }
          }
        }
      }
    }
  }
}

Sales Orders

query SalesOrders {
  x3Sales {
    salesOrder {
      query(first: 25) {
        edges {
          node {
            code {
              orderNumber
              orderDate
              customer { code companyName1 }
              totalAmount
              status
              lines {
                product { code localizedDescription1 }
                quantity
                unitPrice
                lineAmount
              }
            }
          }
        }
      }
    }
  }
}

OData-style Filtering

# Active South African customers
query(first: 50, filter: "country.code eq 'ZA' and isCustomer eq true")

# Orders this year
query(first: 100, filter: "orderDate ge '2026-01-01'")

REPORTING PATTERNS

Sage X3 GraphQL doesn't support aggregation — fetch data and aggregate client-side.

Revenue by Customer

query RevenueReport {
  x3Sales {
    salesOrder {
      query(first: 1000, filter: "orderDate ge '2026-01-01'") {
        edges {
          node {
            code {
              customer { code companyName1 }
              totalAmount
              currency { code }
              orderDate
            }
          }
        }
      }
    }
  }
}
// Then group by customer.code, sum totalAmount

Aged Debtors

query AgedDebtors {
  x3Sales {
    salesInvoice {
      query(first: 500, filter: "outstandingAmount gt 0") {
        edges {
          node {
            code {
              invoiceNumber
              invoiceDate
              dueDate
              customer { code companyName1 }
              totalAmount
              outstandingAmount
            }
          }
        }
      }
    }
  }
}
// Age buckets: Current, 30d, 60d, 90d+

Inventory Alerts

Fetch stock levels and filter where quantityOnHand < reorderPoint.

AI INTEGRATION

The power pattern: natural language in, formatted report out.

Cloudflare Worker Proxy

async function handleGraphQL(request, env) {
  const { query, variables } = await request.json();
  const credentials = btoa(
    `${env.SAGE_X3_USERNAME}:${env.SAGE_X3_PASSWORD}`
  );

  const response = await fetch(env.SAGE_X3_ENDPOINT, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      'Authorization': `Basic ${credentials}`,
    },
    body: JSON.stringify({ query, variables }),
  });

  return new Response(response.body, {
    status: response.status,
    headers: { 'Content-Type': 'application/json' },
  });
}

Natural Language System Prompt

You are a Sage X3 GraphQL expert. Given a natural
language question, generate the appropriate GraphQL query.
The API uses Relay pagination (edges/node) with domains:
x3MasterData, x3Sales, x3Purchasing, x3Stock, x3Accounting.
Always include the `first` parameter for pagination.
Respond with ONLY the GraphQL query.

INSTALL

Works with Claude Code, Cursor, Windsurf, Cline, and 30+ other AI agents:

npx skills add imbilawork/2nth-skills@sagex3-ai

Or install globally with no prompts:

npx skills add imbilawork/2nth-skills@sagex3-ai -g -y

The skill is loaded automatically when your agent works on Sage X3 related tasks.

GOTCHAS