Sf Config Debt Workshop

SF Config Debt Workshop

Run a full client-facing config debt workshop for SAP SuccessFactors Employee Central. From raw metadata to board-ready deliverables in one workflow.

When to Use

• A client asks "is our SF configuration healthy?"
• Pre-migration readiness assessment
• Post-go-live configuration audit
• Building a governance business case for an HR transformation project
• Preparing for a client workshop where you need to demonstrate config debt risk

Prerequisites

• SF Config Debt Scanner MCP server running (connects via mcp__sf_config_debt_scanner__* tools)
• Either: access to SF tenant (API credentials), OR a $metadata XML export file
• No employee data is extracted -- schema and counts only

Evidence and Human Control

• Capture source evidence with tenant, scope and snapshot date.
• Separate observed configuration evidence from assumptions and stakeholder statements.
• No employee data is required or permitted in the deliverable.
• Remediation needs named human approval from the HRIS/SF owner; the skill never writes configuration.
• Re-run the scan after remediation and retain before/after evidence.

Workflow

Step 1: Discovery

Start with assessment questions:

mcp__sf_config_debt_scanner__sf_assessment_questions(category="all")

This returns categorised discovery questions covering governance, custom fields, MDF, picklists, event reasons, foundation objects, and business rules. Use these as the client workshop opening.

Step 2: Scan (Choose Path)

Path A -- Live tenant scan:

mcp__sf_config_debt_scanner__sf_test_connection(base_url="https://apiXX.sapsf.eu")

Confirm connectivity, then:

mcp__sf_config_debt_scanner__sf_scan_tenant(base_url="https://apiXX.sapsf.eu", auth_method="basic", username="...", password="...")

Path B -- Offline XML scan (no tenant access needed):

mcp__sf_config_debt_scanner__sf_scan_metadata_xml(xml_text="...")

Step 3: Score Findings

Rate the findings automatically:

mcp__sf_config_debt_scanner__sf_rate_findings(findings_json="...")

Returns: overall debt score, area breakdown, prioritised 90-day roadmap.

Step 4: Package Output

Produce the workshop pack:

1. Discovery question bank (from Step 1)
2. Findings with severity (from Step 2/3)
3. Executive summary -- top 3 business risks in plain language
4. 90-day phased roadmap (from sf_rate_findings)
5. Next-steps decision tree -- what the client should do on Monday morning

Package in client-safe language. Flag internal-only observations separately.

Edge Cases

• Tenant unreachable: fall back to XML path; note in output that counts could not be verified live
• Truncated XML: partial scan still produces findings on visible entities
• Empty tenant: produce a clean-baseline report stating no debt found
• Very large tenant: increase MCP timeout to 300s; sf_scan_tenant handles batching
• OAuth2 auth: set auth_method="oauth2" and provide client_id, client_secret, company_id
• Custom field limits: pass high_field_limit, custom_field_limit parameters to adjust thresholds per client

Common Pitfalls

1. Sharing internal findings with client: Rate findings output contains unvarnished severity labels. Sanitise before presenting to non-technical audiences. Flag internal sections with [INTERNAL ONLY].
2. Forgetting to test connection first: Use sf_test_connection before sf_scan_tenant -- it's faster to fail early if credentials are wrong.
3. Trusting old XML exports: $metadata changes over time. Note the export date in output.
4. Thresholds too aggressive for small tenants: A company with 200 employees doesn't need the same limits as one with 50,000. Adjust custom_field_limit and high_field_limit.
5. Missing the business impact translation: Don't deliver raw field counts. Translate to business risk: "blank cost center fields → incorrect payroll cost allocation → finance reconciliation errors".

Verification Checklist

• Connection tested (live) or XML validated (offline)
• Findings scored and sorted by severity
• Roadmap includes concrete actions with effort estimates
• Executive summary uses business language, not technical jargon
• Internal-only observations flagged
• Client document has date, tenant reference, and confidence score