finstat-enrichment

name

finstat-enrichment

description

Enrich CRZ contract suppliers with financial data from FinStat.sk — revenue, profit, equity, tax reliability, founding date — and evaluate financial zlte stopy. Use this skill whenever you need to check a supplier's financial health, when critical-validation marks a finding as CONFIRMED or INCONCLUSIVE and recommends FinStat enrichment, when the crz-investigate orchestrator enters the enrichment phase, or when the user asks about a company's financials, revenue, equity, or tax status. Two modes — batch pipeline for many suppliers, manual curl for 1-5 specific companies.

FinStat Supplier Financial Enrichment

Enrich CRZ suppliers with financial data from FinStat.sk and flag financial zlte stopy. This is the "follow the money" step — after sql-analytics finds anomalies and critical-validation confirms them, finstat-enrichment checks whether the supplier is financially healthy enough to deliver.

Input

Field	Description	Example
mode	batch or manual	manual
icos	List of ICOs to enrich	51671824, 44965257
date_from	Period start (for batch)	2026-01-01
date_to	Period end (for batch)	2026-02-01
context	Why these suppliers matter	CONFIRMED double debtor from critical-validation

If the user provides specific ICOs (1-5 companies), use manual mode. If they ask to enrich "all suppliers" or "top suppliers" for a period, use batch mode.

Mode 1: Manual lookup (1-5 companies)

Fetch all companies in parallel — launch all curl/WebFetch calls in the same turn since they're independent. Use curl (not Playwright) for speed.

curl -s "https://finstat.sk/{ico}" -H "User-Agent: Mozilla/5.0"

Where to find data on the page (saves you exploratory calls):

Data point	Location in HTML
Company name	<title>{name} - zisk, tržby...</title>
Trzby, Zisk, Aktiva	og:description meta tag: "Zisk: X, Tržby: Y, Aktíva: Z"
Vlastny kapital	Text near "Vlastný kapitál" followed by >NNN EUR
Datum vzniku	Text near "Dátum vzniku" in a <span>
Celkova zadlzenost	Text near "Celková zadlženosť" followed by N,N %

Also check the local database registers for all ICOs in one query:

SELECT 'vszp' as register, ico, dlh as value FROM vszp_debtors WHERE ico IN ('{ico1}','{ico2}')
UNION ALL SELECT 'socpoist', ico, dlh FROM socpoist_debtors WHERE ico IN ('{ico1}','{ico2}')
UNION ALL SELECT 'fs', ico, dlh FROM fs_tax_debtors WHERE ico IN ('{ico1}','{ico2}')
UNION ALL SELECT 'ruz', ico, CAST(vlastne_imanie AS TEXT) FROM ruz_equity WHERE ico IN ('{ico1}','{ico2}')
UNION ALL SELECT 'tax_rel', ico, status FROM tax_reliability WHERE ico IN ('{ico1}','{ico2}')

Skip state institutions — ministries, universities, state funds won't have finstat.sk profiles. Focus enrichment on private suppliers (s.r.o., a.s.).

Mode 2: Batch pipeline (many suppliers)

Run the enrichment pipeline:

# Top suppliers for a month
uv run python pipeline/enrich_suppliers.py --month 2026-01 --limit 100

# Date range
uv run python pipeline/enrich_suppliers.py --from 2026-01 --to 2026-03 --limit 200

# Single company via pipeline
uv run python pipeline/enrich_suppliers.py --ico 51671824

# Re-evaluate flags only (no scraping)
uv run python pipeline/enrich_suppliers.py --flags-only --month 2026-01

Results are stored in supplier_financials and supplier_red_flags tables.

Query results after pipeline run:

-- Top flagged suppliers
SELECT sf.ico, sf.nazov, sf.trzby, sf.zisk, sf.vlastny_kapital,
       count(DISTINCT srf.flag_type) as flag_types,
       string_agg(DISTINCT srf.flag_type, ',') as flags
FROM supplier_red_flags srf
JOIN supplier_financials sf ON srf.ico = sf.ico
GROUP BY srf.ico
ORDER BY count(DISTINCT srf.flag_type) DESC LIMIT 20;

Zlte stopy rules

Evaluate these 7 rules for each supplier. Apply them whether using manual or batch mode:

Zlta stopa	Severity	Condition	What it means
contract_exceeds_2x_revenue	DANGER	Contract > 2x annual revenue	Firm too small for this contract
negative_equity	DANGER	Vlastne imanie < 0	Firm is technically insolvent
severe_loss	DANGER	Loss > 30% of revenue	Firm bleeding money
tax_unreliable	DANGER	Tax reliability: menej spolahlivy	Tax authority considers them unreliable
missing_rpvs	WARNING	Not in RPVS, contract > 100K	Should be registered as public sector partner
unusually_high_profit	WARNING	Profit > 50% of revenue	Unusual margin, possible shell company
young_company	WARNING	Contract < 1 year after founding	New firm winning big contracts

Framework contract adjustment: For ramcove zmluvy (framework agreements) with multi-year duration, divide the total contract value by the number of years before comparing to annual revenue. A 7-year framework for 145M EUR is ~21M/year — compare that to revenue, not the full 145M.

Risk assessment rubric

Use this rubric for the Hodnotenie section — it standardizes the assessment across suppliers:

Signal	Risk level	Interpretation
Contract < 1x revenue, positive equity	LOW	Firm can handle it from normal operations
Contract 1-3x revenue, positive equity	MEDIUM	Stretched but feasible, may need financing
Contract > 3x revenue OR negative equity	HIGH	Serious capacity concern
Contract > 5x revenue AND negative equity	CRITICAL	Very unlikely to deliver without external help

For multi-year contracts, use the annualized value in this comparison.

Output template

# FinStat enrichment: {context}

## Sumar
- Pocet dodavatelov: {N}
- Financne zlte stopy: {X} DANGER, {Y} WARNING
- Najkritickejsi dodavatel: {name} ({reasons})

---

## Dodavatel 1: {company_name} (ICO {ico})

| Ukazovatel | Hodnota |
|---|---|
| Trzby (revenue) | {trzby} EUR |
| Zisk (profit) | {zisk} EUR |
| Vlastny kapital (equity) | {vlastny_kapital} EUR |
| Celkova zadlzenost | {zadlzenost}% |
| Datum vzniku | {datum} |
| Danova spolahlivost | {tax_status} |

### Zmluvy v CRZ
| ID | Suma | Nazov zmluvy |
|---|---|---|
| {id} | {suma} EUR | {nazov} |

### Financne zlte stopy
| Zlta stopa | Severity | Detail |
|---|---|---|
| {flag_type} | {severity} | {detail} |

### Hodnotenie
**Riziko: {LOW/MEDIUM/HIGH/CRITICAL}** — {1-2 sentence assessment}

---

[...repeat for each supplier...]

---

**Enrichment ukonceny.** {summary of most concerning findings}