Agent-authored patterns — v1 MCP CSV (dotted field IDs)

Use this file for reports produced by allv1_AdsApiv1CreateReport and downloaded via MCP. CSV headers are exactly the v1 field IDs from mcp-workflow.md (e.g. searchTerm.value, metric.totalCost). Do not use legacy Advertising Console column names (Customer Search Term, 7 Day Total Sales, …) — those are a different export.

Validate expected headers

import csv

with open(csv_path, encoding="utf-8") as f:
    reader = csv.DictReader(f)
    rows = list(reader)

REQUIRED_V1 = {
    "date.value",
    "searchTerm.value",
    "metric.totalCost",
    "metric.sales",
    "metric.clicks",
    "metric.impressions",
    "metric.purchases",
    "budgetCurrency.value",
}
if not rows:
    raise ValueError("Empty CSV")
missing = REQUIRED_V1 - set(rows[0].keys())
if missing:
    raise ValueError(f"Missing v1 columns: {missing}")

Partition PAT ASINs vs customer queries (searchTerm.value)

import re

ASIN = re.compile(r"^B0[A-Z0-9]{8}$")

def is_asin_shaped(v: str) -> bool:
    return bool(ASIN.match(str(v).strip().upper()))

customer_rows = [r for r in rows if not is_asin_shaped(r["searchTerm.value"])]
asin_rows = [r for r in rows if is_asin_shaped(r["searchTerm.value"])]

Run gold / rising / keyword negatives on customer_rows only. Summarize asin_rows under product-targeting / PAT (not keyword negatives).

Aggregate by search term (v1 metrics)

v1 search-term reports are typically multi-row per searchTerm.value (date × campaign × ad group). Always aggregate before segment rules.

from collections import defaultdict

agg = defaultdict(
    lambda: {
        "metric.totalCost": 0.0,
        "metric.sales": 0.0,
        "metric.clicks": 0,
        "metric.impressions": 0,
        "metric.purchases": 0,
    }
)

for r in customer_rows:
    t = r["searchTerm.value"]
    agg[t]["metric.totalCost"] += float(r["metric.totalCost"] or 0)
    agg[t]["metric.sales"] += float(r["metric.sales"] or 0)
    agg[t]["metric.clicks"] += int(float(r["metric.clicks"] or 0))
    agg[t]["metric.impressions"] += int(float(r["metric.impressions"] or 0))
    agg[t]["metric.purchases"] += int(float(r["metric.purchases"] or 0))

# Derived (after aggregation):
# acos = cost / sales if sales > 0 else None
# cvr = orders / clicks if clicks > 0 else None

For brand/generic regex and threshold application, use the sibling references in this skill (agent-codegen-patterns.md Patterns 3–4, segment-rules.md, threshold-elicitation.md) — those are format-agnostic; only the column keys here are v1-specific.

Anti-patterns (v1)

• Treating metric.cost, metric.spend, metric.spendAmount as cost — rejected by the v1 endpoint; use metric.totalCost.
• Expecting metric.sales7d, metric.purchases7d, or other suffixed attribution fields on this CreateReport path — not supported here; use metric.sales / metric.purchases and disclose default attribution.
• Expecting keyword.text, keyword.value, keyword.matchType in the export — not available as v1 output fields in current catalog for this pattern.
• Renaming v1 headers to legacy “friendly” names before aggregation — breaks the contract and duplicates mapping bugs.