Discover Billing

Execute ALL steps in order. Do not skip or optimize.

Step 0: Self-Scan for Billing Files

Scan the target directory for billing data:

• **/*billing*.csv — GCP billing export CSV
• **/*billing*.json — BigQuery billing export JSON
• **/*cost*.csv, **/*cost*.json — Cost report exports
• **/*usage*.csv, **/*usage*.json — Usage report exports

Exclude .migration/** — migration run artifacts (e.g. openai-usage-profile.json, openai-capture/) are never billing input.

Exit gate: If NO billing files are found, exit cleanly. Return no output artifacts. Other sub-discovery files may still produce artifacts.

Step 1: Parse Billing Data

Supported formats:

• GCP billing export CSV
• BigQuery billing export JSON

Extract from each line item:

• service_description — GCP service name
• sku_description — Specific SKU/resource
• cost — Cost amount
• usage_amount — Usage quantity
• usage_unit — Usage unit (e.g., hours, bytes, requests)

Group by service and calculate monthly totals.

Step 1.5: Identify Commitments and Discounts

Scan billing line items for GCP Committed Use Discount (CUD) artifacts and other billing-level discounts. These are not workload costs — they are financial instruments that must be separated from actual resource usage.

Detection patterns

Additional signals:

• resourceGlobalName contains project_commitments → commitment fee row
• resourceName starts with commitment- → commitment fee row

Extraction

For each detected commitment fee row, extract:

• Term: Parse from SKU (e.g., “1 Year” or “3 Year”)
• Covered service: Parse from SKU (e.g., “E2 Cpu”, “E2 Ram”, “GCE”, “Cloud SQL”)
• Region: From the region field
• Monthly fee: Sum costAtListUSD for all rows in that commitment

For discount credits, sum per type across all services:

• Total committedUsageDiscount (negative values = credits applied)
• Total sustainedUsageDiscount
• Total freeTier

Cost basis determination

If any CUD-related columns or rows are detected:

• Use costAtListUSD (list price) as the baseline for all service costs in Step 2
• Record total_at_list (sum of all costAtListUSD)
• Record total_net_of_discounts (list price + all discount credits)
• Calculate effective_discount_percent: (total_at_list - total_net_of_discounts) / total_at_list × 100

If no CUD columns are present in the export (older export format), use the available cost column as-is and set cost_basis.uses_list_price to false.

Step 2: Build Service Usage Profile

From the parsed billing data:

1. List all GCP services with non-zero spend
2. Calculate monthly cost per service using list price (costAtListUSD) when available — exclude commitment fee rows from service totals
3. Identify top services by spend (sorted descending)
4. Note usage patterns (consistent vs bursty spend)

Step 3: Flag AI Signals

Scan billing line items for AI-relevant patterns. For each match, record the pattern, line item details, and confidence score.

Step 4: Generate billing-profile.json

Write $MIGRATION_DIR/billing-profile.json with the following structure:

Load references/shared/schema-discover-billing.md and validate the output against the billing-profile.json schema.

After generating the output file, the parent discover.md handles the phase status update — do not update .phase-status.json here.

Scope Boundary

This phase covers Discover & Analysis ONLY.

FORBIDDEN — Do NOT include ANY of:

• AWS service names, recommendations, or equivalents
• Migration strategies, phases, or timelines
• Terraform generation for AWS
• Cost estimates or comparisons
• Effort estimates

Your ONLY job: Inventory what exists in GCP. Nothing else.