Unified API Cross-Reference Table

Context#

SAP Digital Manufacturing exposes its APIs through two separate sources: the service registry (where APIs are registered by ID and path for use in production processes) and the SAP Business Accelerator Hub (where the detailed API documentation, schemas, and try-out pages live). The two aren’t a one-to-one mapping—some APIs exist only in the service registry, such as internal or edge-specific ones, while others appear only on the Hub. With no consolidated view and inconsistent naming conventions between the sources, users had no reliable way to move from an API they encountered in one place to its documentation in the other, and a number of endpoints were effectively invisible.

My Role#

I built generate-api-table, a Claude Code plugin that automatically generates a unified DITA XML cross-reference table mapping every service-registry API to its corresponding SAP Business Accelerator Hub endpoint. I designed a dual-strategy matcher—a forward pass from the registry config to the Hub artifacts and a reverse pass that discovers APIs present in the artifacts but missed on the way forward—so the table captures endpoints that manual lookup would never surface. I also presented the approach and a live demo at the UA team meeting to show how AI automation can eliminate manual API-lookup work.

Key Challenges#

• API information was split across two systems—the service registry (config.yaml) and the SAP Business Accelerator Hub—with no consolidated inventory linking them.
• Matching was not straightforward: the two sources use inconsistent naming, so a single exact match was never enough. The forward pass relies on seven matching strategies (exact operationId, stripped operationId, summary, external-path variants, direct path, case-insensitive operationId, and fuzzy summary prefix).
• Some APIs existed only in the artifacts and were previously invisible; a reverse-matching pass was needed to detect them programmatically rather than by hand.
• The output had to be accurate and complete enough to publish directly into the DITA CCMS XML editor, which required automated cross-validation—completeness, exact name/path match to source, no duplicate IDs, valid Hub URLs, and correct column structure—plus deduplication that still preserves genuine v1/v2 API pairs.

Deliverables#

• A Claude Code plugin (skill + Python scripts) that generates a five-column DITA XML table—Display Name, ID, URL/Path, SAP Business Accelerator Hub Endpoint, and Hub URL—ready for direct copy-paste into the CCMS XML editor.
• A dual forward-plus-reverse matching engine that both maps known APIs and discovers previously unmatched endpoints, with deduplication by Hub URL.
• A validation script that verifies completeness and accuracy against the source config before publishing.
• The published outcome: “API Cross-Reference: Service Registry and SAP Business Accelerator Hub” on SAP Help Portal—a single searchable, sortable table that lets users cross-reference APIs from either direction, jumping from a registry path straight to the Hub documentation, or from the Hub back to the exact registry path and ID to call in a production process.

Impact & Business Value#

• Reduced the time it takes users to find an API and navigate to its documentation, replacing manual cross-system lookup with one searchable table.
• Enabled users to look up any API from either direction, regardless of where or how they first encountered it.
• Improved API discoverability with direct links to specifications, schemas, and try-out pages on the Hub.
• Surfaced endpoints that were previously undiscoverable, making the API landscape more complete and auditable.
• Demonstrated a practical, repeatable application of AI automation in the documentation workflow for the wider UA team.

Reference: API Cross-Reference: Service Registry and SAP Business Accelerator Hub — SAP Help Portal ↗