Field Name Translation

The backend (Python / FastAPI) uses snake_case for all field names internally. API response models inherit from CamelModel (i4g.api.camel), which uses Pydantic's alias_generator = to_camel to serialise JSON keys as camelCase automatically. No manual translation is needed for response models.

The frontend SDK and UI components use camelCase natively.

Current State

Automatic Translation (via CamelModel)

All Pydantic response models in response_models.py and cases.py inherit from CamelModel. Python code uses snake_case field names; JSON output is camelCase. No manual mapping required.

SDK cleanup (D79): The normalizeVerificationArtifact and normalizeDossierVerification functions have been removed — the backend now returns camelCase for verification endpoints directly.

Remaining Manual Translation Points

These sites still perform explicit mapping because the data flows through raw dicts (Dict[str, Any]) rather than typed CamelModel instances:

#	File	Function(s)	Reason
1	ui/apps/web/src/lib/platform-client.ts	fetchCoreSearch (request body)	Client→server query params
2	ui/packages/sdk/src/index.ts	normalizeDossierRecord, normalizeDownloads	Dossier list returns raw dicts
3	ui/apps/web/src/lib/server/reviews-service.helpers.ts	mapHistoryEvent, mapSavedSearch, mapHybridSearchSchemaPayload	Events/searches are raw store dicts
4	ui/apps/web/src/lib/search-links.ts	buildSearchHref	URL param building (dual-tolerance)

Eliminated Translation Points

The following were removed by D79:

• normalizeVerificationArtifact — backend VerifyArtifact now serialises as camelCase

• normalizeDossierVerification — backend DossierVerifyResponse now serialises as camelCase

• detokenizeResponseSchema snake_case keys — removed (endpoint retired)

• caseDetailSchema graph_nodes/graph_links — updated to graphNodes/graphLinks

Guidelines for New Endpoints

1. Response models: Inherit from CamelModel — camelCase is automatic.

2. Raw dict responses (e.g., List[Dict[str, Any]]): Keys pass through unchanged. If the frontend needs camelCase, add a normalize function.

3. Request bodies: Currently snake_case. Frontend sends snake_case; Pydantic accepts both casings when populate_by_name=True.

4. Query parameters: Defined as function parameters — always snake_case in the URL. Frontend maps camelCase values to snake_case params explicitly.

Intelligence & Export Models (Sprint 2)

The intelligence and export endpoints follow the same CamelModel convention. Key field mappings:

Python Field (snake_case)	JSON / SDK Field (camelCase)	Model
entity_type	entityType	Entity list/detail
canonical_value	canonicalValue	Entity list/detail
case_count	caseCount	Entity stats
loss_sum	lossSum	Entity stats
first_seen	firstSeen	Entity / Indicator
last_seen	lastSeen	Entity / Indicator
indicator_value	indicatorValue	Indicator
confidence_score	confidenceScore	Indicator
campaign_name	campaignName	Campaign stats
risk_score	riskScore	Campaign stats
total_entities	totalEntities	Platform KPIs
total_indicators	totalIndicators	Platform KPIs
active_campaigns	activeCampaigns	Platform KPIs

No manual translation is needed — all models inherit from CamelModel.