Measure supplier dependency

# Supplier concentration and dependency

## Purpose

Measure how much spend rests on a few suppliers: how many account for 80%, which ones weigh more than 10% each, and which spend categories depend on a single supplier.

State the exact period (the last 12 months by default; 366 days at most per call).

## Step 1 — Spend per supplier

```json
{
  "tool": "spendesk_analyze_spend",
  "params": { "fromDate": "YYYY-MM-DD", "toDate": "YYYY-MM-DD", "groupBy": ["supplier"], "limit": 50 }
}
```

⚠️ Three points verified live:

- **By default, only 10 rows come back**; the rest is rolled into an "Others" row. Pass `limit: 50`, the maximum.
- Beyond 50 suppliers, the "Others" row remains: the 80% calculation stays exact as long as the threshold is reached within the first 50. Otherwise, say that concentration is low and the detail is not available.
- **Do not reuse `share`**: it does not match the share of the net total as soon as there are credit notes. Recompute each share on `summary.totalAmount`.

Credit notes are included and netted off. For concentration on gross purchases, add `"filters": { "payableTypes": [...] }` without `creditNote`.

## Step 2 — Pareto

From the rows sorted by amount:

- number of suppliers that add up to 80% of `summary.totalAmount`;
- suppliers weighing more than 10% each: critical dependency;
- the top supplier's share.

## Step 3 — Categories that depend on one supplier

There is no "category" dimension: expense categories come out under `analyticalCode`, **mixed with the values of other analytical fields**. Isolate them:

```json
{ "tool": "get_expense_categories", "params": {} }
```

Then:

```json
{
  "tool": "spendesk_analyze_spend",
  "params": { "fromDate": "YYYY-MM-DD", "toDate": "YYYY-MM-DD", "groupBy": ["analyticalCode", "supplier"], "limit": 50 }
}
```

Keep only the rows whose `id` is a category ID. In each category's `breakdown`, flag those where one supplier exceeds 90%: that is a single-supplier dependency.

## Step 4 — Records of critical suppliers

```json
{ "tool": "get_suppliers", "params": { "ids": ["<id1>", "<id2>"] } }
```

Read `address.country` and `bankCountry`. An empty `iban` does not prove there are no bank details (non-IBAN accounts are never returned): do not treat it as a risk signal on its own. Judging country risk is the user's call.

## Report

```
Supplier concentration — from <date> to <date>
Total spend: X EUR

N suppliers account for 80% of spend
Suppliers above 10%: name, amount, share
Categories depending on one supplier: category, supplier, share
```

## Limits to state

- Only spend that went through Spendesk is counted.
- A dominant supplier is not necessarily a risk if it is a chosen partner: the skill measures, it does not judge.
- To see whether concentration is rising, run the analysis again on the previous 12 months.

## Observed result (demo account, 5 October 2026)

Over 12 months and EUR 113,720.26 of spend, the top supplier weighed EUR 60,000.00, i.e. 52.8% of the total, and three suppliers each exceeded 10%. The share shown by the tool for the top one (51.6%) differed from the share recomputed on the total.

## Example prompts
- "How many suppliers account for 80% of our purchases?"
- "Which suppliers are we most dependent on?"
- "Are there categories where we only have one supplier?"