salesforce-field-mapping-skill

salesforce-field-mapping-skill

Pure-generation T0 skill that maps CSV/spreadsheet column headers to Salesforce field API names with type mismatch detection. The daily-driver for admins migrating from HubSpot, Pipedrive, Excel exports, and legacy CRMs. Outputs a complete mapping table, type mismatch report, missing-field list, and a Data Loader–ready column header row.

When This Skill Owns the Task

Use salesforce-field-mapping-skill when the work is to design or validate a field mapping for a data import:

• "I'm migrating from HubSpot — map these column headers to Salesforce Contact fields"
• "Here's my Excel export. What are the Salesforce API names?"
• "I need a Data Loader field mapping CSV for importing Opportunities from Pipedrive"
• "Flag any type mismatches in my column list before I import"
• "Which of my CSV columns don't have matching Salesforce fields?"

Delegate elsewhere when:

Required Context to Gather First

Before generating a mapping, confirm:

1. Target Salesforce object — standard (Contact, Lead, Account, Opportunity, Case) or custom (Custom_Object__c).
2. Source system — HubSpot, Pipedrive, Salesforce Classic, Excel/CSV export, Marketo, or other. Source system determines default field label conventions.
3. Column header list — the complete list of CSV/spreadsheet column headers, either as a comma-separated list or pasted table.
4. Custom fields in scope — list any known custom fields on the target object that may receive data (e.g., HubSpot_Contact_ID__c).
5. Operation type — insert, update, or upsert. Affects whether Id or an external ID field is required in the mapping.
6. External ID field — for upsert: which field is the external ID key? (e.g., HubSpot_Contact_ID__c, Email)
7. Multi-value fields — any source columns that contain comma-separated multi-values intended for multi-select picklists.

If the column header list is not provided, ask before proceeding.

Recommended Workflow

Step 1 — Normalize source column headers

Apply label normalization to make headers comparable:

• Strip leading/trailing whitespace
• Collapse multiple spaces to single space
• Remove non-alphanumeric prefix/suffix characters
• Flag headers containing special characters that may corrupt Data Loader (e.g., #, $, %, (, ), /)

Step 2 — Derive candidate API names from labels

For each normalized column header, apply the API name derivation rules (see references/api-name-normalization.md):

1. Replace spaces with underscores
2. Remove characters not in [A-Za-z0-9_]
3. Strip leading digits (Salesforce API names must start with a letter)
4. Collapse consecutive underscores to single underscore
5. Truncate to 40 characters (API name limit)
6. Append __c for custom fields
7. Check against reserved words list

Step 3 — Match against known standard field API names

Apply fuzzy matching against the standard object's well-known fields:

For Contact: FirstName, LastName, Email, Phone, MobilePhone, Title, Department, AccountId, MailingStreet, MailingCity, MailingState, MailingPostalCode, MailingCountry, LeadSource, Salutation, Birthdate, Description, DoNotCall, HasOptedOutOfEmail, OwnerId, ReportsToId

For Lead: FirstName, LastName, Company, Email, Phone, MobilePhone, Status, LeadSource, Title, Industry, AnnualRevenue, NumberOfEmployees, Rating, Street, City, State, PostalCode, Country, Description, HasOptedOutOfEmail, DoNotCall

For Account: Name, Type, Industry, AnnualRevenue, NumberOfEmployees, BillingStreet, BillingCity, BillingState, BillingPostalCode, BillingCountry, Phone, Website, Description, OwnerId, ParentId

For Opportunity: Name, StageName, CloseDate, Amount, Probability, AccountId, OwnerId, Type, LeadSource, Description, ForecastCategory, ForecastCategoryName, CampaignId, NextStep

Step 4 — Detect type mismatches

For each mapped pair, compare source data type to Salesforce field type. Flag mismatches (see references/type-mismatch-detection.md):

Step 5 — Flag missing and unmappable fields

Produce two lists:

1. Unmapped source columns — columns with no clear Salesforce target
2. Required fields missing from source — Salesforce required fields (e.g., LastName on Contact, LastName + Company on Lead, Name + StageName + CloseDate on Opportunity, Name on Account) not present in the source column list

Step 6 — Normalize picklist values

For any column mapping to a standard picklist field, list the expected Salesforce API values and flag mismatches:

• LeadSource: Cold Call, Web, Phone Inquiry, Partner Referral, Purchased List, Other, Word of mouth, Employee, Internal, Partner
• StageName: Prospecting, Qualification, Needs Analysis, Value Proposition, Id. Decision Makers, Perception Analysis, Proposal/Price Quote, Negotiation/Review, Closed Won, Closed Lost
• Industry: Agriculture, Apparel, Banking, Biotechnology, Chemicals, Communications, Construction, Consulting, Education, Electronics, Energy, Engineering, Entertainment, Environmental, Finance, Food & Beverage, Government, Healthcare, Hospitality, Insurance, Machinery, Manufacturing, Media, Not For Profit, Recreation, Retail, Shipping, Technology, Telecommunications, Transportation, Utilities, Other

Step 7 — Detect special character and collision risks

• Flag API names that would collide (two source columns normalizing to the same API name)
• Flag reserved words: Id, Name, OwnerId, CreatedDate, LastModifiedDate, SystemModstamp, IsDeleted — these cannot be used as custom field API names
• Flag API names exceeding 40 characters after normalization

Step 8 — Emit structured output

Produce the full output block per the Output Format section below, including a Data Loader–ready header row.

Quality Scoring Rubric (100-point)

Score every output before emitting. Threshold: 85+ ship, 70–84 ship with caveat, below 70 reject and revise.

Scoring penalties:

• Missing a HIGH-risk type mismatch: -15 per missed mismatch
• Missing a required field from target object: -10
• Incorrect standard API name: -15 per error
• Multi-select picklist mapped to single picklist without flag: -20
• Lookup/relationship field mapped without upsert key guidance: -10

T0 Contract

This skill is static-review (T0) only:

• No Salesforce org connection required or permitted.
• No CLI commands are executed.
• No live field metadata is fetched.
• The mapping is generated from known standard field sets and the user's provided column list.
• If the target object has custom fields that the user has not listed, mappings to those custom fields cannot be generated here — use salesforce-metadata-fetcher-skill to fetch them first.

Refusal Triggers

Stop and decline if:

• The request is to execute a live data migration (route to salesforce-bulk-data-ops-skill).
• The target object is not specified after two clarification attempts.
• The column header list is not provided after one prompt.
• The operation involves permanent hard-deletes (T3 prohibited — route to salesforce-live-guard-agent).

Output Format

verdict: "ship | ship-with-caveat | reject"
quality_score: <0-100>
quality_notes: "<what drove the score>"

mapping_summary:
  source_system: "<HubSpot | Pipedrive | Excel | Salesforce Classic | Other>"
  target_object: "<SObjectApiName>"
  operation: "<insert | update | upsert>"
  external_id_field: "<FieldApiName | none>"
  total_source_columns: <integer>
  mapped_columns: <integer>
  unmapped_columns: <integer>
  type_mismatch_count: <integer>
  required_fields_missing: <integer>

field_mapping:
  - source_column: "<source header>"
    salesforce_api_name: "<FieldApiName>"
    salesforce_field_type: "<Text|Date|Number|Currency|Picklist|Lookup|Checkbox|etc>"
    confidence: "<high|medium|low>"
    type_mismatch: "<none|HIGH|MEDIUM|LOW>"
    mismatch_detail: "<description of mismatch if present>"
    transform_recommended: "<none|DATEVALUE|VALUE|Picklist normalization|External ID lookup>"
    notes: "<any special handling>"

unmapped_source_columns:
  - column: "<source header>"
    reason: "<no standard match | ambiguous | custom field needed>"
    suggested_custom_api_name: "<Suggested_Name__c>"

required_fields_missing_from_source:
  - field: "<FieldApiName>"
    required_for: "<insert|update|upsert>"
    impact: "<import will fail | data quality risk>"

picklist_value_mismatches:
  - source_column: "<source header>"
    salesforce_field: "<FieldApiName>"
    source_values_found: ["<val1>", "<val2>"]
    valid_salesforce_values: ["<val1>", "<val2>"]
    mismatched_values: ["<val1>"]
    normalization_recommended: "<map 'val1' → 'Correct Value'"

collision_and_special_char_warnings:
  - column: "<source header>"
    issue: "<special chars | api name collision | reserved word | length overflow>"
    recommendation: "<how to resolve>"

data_loader_header_row: "<Comma-separated list of Salesforce API names in source column order>"

handoff_suggestions:
  - "<e.g., fetch custom fields from org with salesforce-metadata-fetcher-skill>"
  - "<e.g., run data migration with salesforce-bulk-data-ops-skill>"

assumptions:
  - "<explicit list of assumptions made>"

Redaction Rules

This is a T0 generation skill with no live org connection. Apply:

1. Never emit actual production org IDs, user IDs, or credentials even if the user includes them in their prompt.
2. If the user pastes sample data rows alongside headers, do not include actual record values (email addresses, names, phone numbers) in the output.
3. Sanitize pasted column lists that contain Org IDs (18-char starting 00D) before incorporating into output.

Handoff Rules

Stop Conditions

• Import will fail because required fields (LastName, Company, StageName + CloseDate + Name) are entirely missing from the source — stop, surface the list, require resolution before producing a mapping.
• Every source column maps to the same single Salesforce field (e.g., all columns appear to be "Name") — stop and request clarification.
• Requested mapping includes hard-delete or permanent-destroy operations — stop and route to salesforce-live-guard-agent.

Security Notes

• T0 static generation only: No org connection, no CLI execution, no MCP calls.
• No credential handling: Skill never requests or emits OAuth tokens, session IDs, or Connected App credentials.
• PII in sample data: If user pastes real data rows, do not include field values containing PII in the output mapping table. Process headers only.
• Upsert external ID guidance: External ID fields used in upsert operations must have "External ID" checked on the field definition in Salesforce Setup — flag this requirement explicitly.

Reference File Index