Agricultural Commodity Climate MCP Server

Agricultural commodity climate risk intelligence for AI agents via the Model Context Protocol. This MCP server gives any AI assistant — Claude, GPT-4, Cursor, or a custom agent — direct access to live weather stress analysis, pest emergence monitoring, trade concentration scoring, and price shock probability for any crop or growing region on Earth.

The server orchestrates 8 public data sources in parallel: NOAA weather alerts, multi-day forecasts, GDACS disaster events, UN COMTRADE trade flows, World Bank agricultural indicators, GBIF biodiversity occurrence records, FRED commodity price series, and Nominatim geocoding. It synthesises these into four quantified scoring models and a composite Commodity Risk Score (0–100) with actionable recommendations. No API keys for downstream sources are required — the server handles all orchestration.

⬇️ What data can you access?

MCP Tools

Why use Agricultural Commodity Climate MCP?

Commodity buyers, traders, underwriters, and food security analysts face the same problem: relevant risk signals are scattered across NOAA, GDACS, COMTRADE, FRED, and GBIF — all in different formats, all requiring separate queries. Building a unified view manually takes hours. By the time the data is assembled, weather conditions have shifted.

This MCP server puts all eight sources into a single tool call. Ask your AI agent "What is the wheat price shock probability given current Kansas growing conditions?" and it runs FRED price analysis, NOAA weather alerts, GDACS disaster checks, and COMTRADE concentration scoring simultaneously, returning a scored result in 30–90 seconds.

• Scheduling — run daily growing region checks and push results to your risk dashboard on Apify Schedules
• API access — trigger tool calls from Python, JavaScript, or any HTTP client via the Apify API
• Proxy infrastructure — all downstream data source calls use Apify's built-in proxy and retry infrastructure
• Monitoring — get Slack or email alerts when weather stress scores exceed thresholds via Apify Webhooks
• Integrations — pipe results to Zapier, Make, Google Sheets, trading systems, or HubSpot

Features

• Weather Stress Index (0–100) — scores growing regions across four components: NOAA severe crop-threat alerts (drought, flood, frost, freeze, heat, hail, tornado, hurricane, wildfire), forecast temperature extremes above 38°C or below freezing, GDACS agricultural-impact disasters, and compound weather stress when multiple factors coincide
• 9 crop-threat alert categories — the alert parser matches against: drought, flood, frost, freeze, heat, hail, tornado, hurricane, and wildfire for targeted crop relevance
• Pest Emergence Score (0–100) — uses GBIF species occurrence records to detect 10 invasive agricultural threat keywords: locust, armyworm, borer, weevil, aphid, rust, blight, fusarium, phytophthora, and whitefly; overlays World Bank cereal yield and agricultural GDP data for vulnerability amplification
• Spread velocity detection — filters GBIF observations from 2024 onward to distinguish active spread from historical records
• Trade Disruption Score with HHI — computes Herfindahl-Hirschman Index from UN COMTRADE partner trade values; HHI above 2,500 triggers concentrated supply chain signal; identifies 2-partner critical dependencies
• FRED price volatility model — detects period-over-period changes above 5% and 15% thresholds; compares recent third of price history against earlier third to identify trend direction
• Price Shock Probability (0–100) — four-component model: FRED price trend analysis, weather/disaster supply risk factor count, trade flow concentration, and compound shock probability when multiple factors align
• Composite Commodity Risk Score — weighted average: weather stress 30%, price shock 30%, trade disruption 25%, pest emergence 15%; override to CRITICAL when crop failure risk and imminent price shock both present
• Five-level verdict scale — LOW_RISK / MANAGEABLE / ELEVATED / HIGH_RISK / CRITICAL with hedge recommendations auto-generated at each threshold
• Parallel data fetching — all upstream actor calls run via Promise.all, reducing 8 sequential calls to concurrent fetch; each sub-actor allocated 512 MB and 120-second timeout
• Per-tool spending limits — each tool checks eventChargeLimitReached before running; never over-spends the budget cap you set at run time
• Standby mode operation — server runs persistently on Apify's standby infrastructure, eliminating cold-start latency for agent workflows

Use cases for agricultural commodity risk analysis

Commodity trading and futures desks

Traders managing wheat, soybeans, corn, or coffee positions need early warning of supply disruptions before they materialize in price. Run price_shock_probability each morning on your active commodity positions. The FRED trend analysis combined with NOAA and GDACS supply disruption signals gives a quantified probability of a price move before exchange open.

Crop insurance underwriting

Underwriters pricing weather-index insurance products need a current Weather Stress Index for each insured region. Call crop_region_risk_assessment to get a scored assessment for a specific growing region, pulling NOAA alerts, multi-day forecasts, and GDACS disaster events into a single 0–100 score. The signals output tells you which specific conditions are driving the risk.

Food company supply chain planning

Procurement teams running just-in-time commodity supply chains need to know their supplier concentration risk before a weather event hits. Use trade_dependency_analysis to compute the HHI concentration score for each commodity. When HHI exceeds 2,500, you have a concentrated dependency that warrants safety stock or alternative supplier qualification.

Agricultural investment portfolio management

Portfolio managers allocating across commodity-linked equities or structured products need comparable risk metrics across different crops. compare_commodity_risks returns the full composite Commodity Risk Score for any commodity, enabling apples-to-apples comparison of wheat vs. soybeans vs. palm oil risk in a single call.

Government and NGO food security monitoring

Food security analysts in international organizations need country-level vulnerability assessments combining trade dependency, agricultural GDP share, and disaster exposure. food_security_vulnerability queries World Bank agricultural indicators, COMTRADE food import data, GDACS recent disasters, and FRED food price series for any country and returns a composite assessment with actionable signals.

Pest and invasive species early warning

Crop protection teams and agricultural extension services can use pest_emergence_alert to detect emerging pest and disease pressure in any growing region. GBIF observation data for locusts, armyworms, rust fungi, Phytophthora, and other crop pathogens is filtered for recent spread velocity and cross-referenced against regional agricultural vulnerability.

How the scoring models work

Weather Stress Index

The scoreWeatherStress function builds a 100-point score across four components. NOAA alerts are scanned for 9 crop-threat event types; each matching alert adds 5 points and each extreme/severe severity rating adds an additional point, capped at 30. Forecast data is parsed for temperature extremes above 38°C (approximate crop heat damage threshold) and below 0°C (frost), with drought indicators triggered when high temperatures exceed 35°C and precipitation is below 1mm; capped at 25. GDACS events are filtered for flood, drought, cyclone, and storm event types and red/orange alert levels; capped at 25. A compound bonus of up to 20 points is awarded when both severe alerts and temperature extremes are present, or when drought indicators and active disasters co-occur. The final score maps to: FAVORABLE (0–19), MILD_STRESS (20–39), MODERATE_STRESS (40–59), SEVERE (60–79), CROP_FAILURE_RISK (80–100).

Pest Emergence Score

scorePestEmergence queries GBIF species occurrence records and scans scientific names against 10 invasive pest/pathogen keywords. Each invasive species detection adds 8 points, capped at 40 via the species component. World Bank indicators are checked for agricultural-GDP-share above 20%, cereal yields below 2,000 kg/ha, and low arable land percentage to score regional agricultural vulnerability; capped at 30. Spread velocity is measured by filtering GBIF records with observation dates from 2024 onward — each recent observation adds 2 points, capped at 20. A 10-point compound bonus applies when invasive species detections and agricultural vulnerability both register. Final scale: CLEAR / LOW / MODERATE / HIGH / OUTBREAK.

Trade Disruption Score and HHI

scoreTradeDisruption extracts trade partner names and values from COMTRADE records, computes each partner's share of total trade value, and calculates the Herfindahl-Hirschman Index as the sum of squared market shares (×10,000). HHI above 2,500 triggers a concentrated supply chain signal. The HHI score is capped at 35 points. FRED price data is then scanned for period-to-period changes exceeding 5% (1 volatility point) and 15% (2 additional points), capped at 30. Trade volume and cascade risk add up to 35 more points. Final scale: DIVERSIFIED / LOW / MODERATE / CONCENTRATED / CRITICAL_DEPENDENCY.

Price Shock Probability

scorePriceShock divides FRED price series into thirds and compares the recent-third average against the earlier-third average. A 15%+ upward trend adds 3 volatile indicators; a 30%+ trend adds 3 more. Supply risk factors are counted by scanning NOAA and GDACS events for 6 crop-threat keywords (drought, flood, frost, freeze, heat, storm). Trade concentration with 3 or fewer partners adds a trade risk multiplier. A compound bonus of up to 15 points is applied when price trends and supply disruption signals co-occur. Final scale: STABLE / MILD / MODERATE / HIGH / IMMINENT.

Composite Commodity Risk Score

generateCommodityRisk combines the four models with fixed weights: weather stress 30%, price shock 30%, trade disruption 25%, pest emergence 15%. There is one hard override rule: when weatherLevel === 'CROP_FAILURE_RISK' and shockLevel === 'IMMINENT' both trigger simultaneously, the verdict is forced to CRITICAL regardless of the weighted score. Auto-generated hedge recommendations are appended when: weather score ≥ 60 (activate crop insurance), pest level = OUTBREAK (source alternative regions), HHI ≥ 2,500 (diversify supplier base), price shock score ≥ 50 (consider futures hedging), composite ≥ 55 (increase inventory buffers).

How to connect this MCP server

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "agricultural-commodity-climate": {
      "url": "https://agricultural-commodity-climate-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Cursor, Windsurf, or Cline

In your MCP settings, add a new server with:

• URL: https://agricultural-commodity-climate-mcp.apify.actor/mcp
• Transport: Streamable HTTP
• Auth: Bearer token (your Apify API token)

Programmatic HTTP call

curl -X POST https://agricultural-commodity-climate-mcp.apify.actor/mcp \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "compare_commodity_risks",
      "arguments": {
        "commodity": "wheat",
        "region": "Kansas"
      }
    },
    "id": 1
  }'

Python (apify-client)

from apify_client import ApifyClient

client = ApifyClient("YOUR_API_TOKEN")

run = client.actor("ryanclinton/agricultural-commodity-climate-mcp").call(run_input={})

# The MCP server runs in standby mode — connect via HTTP after the server starts
# For direct tool calls, use the HTTP endpoint above
print(f"Server running. Connect at: https://agricultural-commodity-climate-mcp.apify.actor/mcp")

Output example

A compare_commodity_risks call for wheat in the Kansas growing region returns:

{
  "commodity": "wheat",
  "compositeScore": 62,
  "verdict": "HIGH_RISK",
  "weatherStress": {
    "score": 71,
    "severeAlerts": 4,
    "extremeTemps": 3,
    "weatherLevel": "SEVERE",
    "signals": [
      "4 severe weather alerts — crop-threatening conditions",
      "3 drought-condition days — water stress risk",
      "3 extreme temperature events — crop damage risk"
    ]
  },
  "pestEmergence": {
    "score": 18,
    "speciesDetected": 7,
    "invasiveCount": 1,
    "pestLevel": "LOW",
    "signals": []
  },
  "tradeDisruption": {
    "score": 55,
    "tradePartners": 3,
    "concentrationHHI": 2840,
    "disruptionLevel": "CONCENTRATED",
    "signals": [
      "Trade HHI 2840 — concentrated commodity supply chain",
      "Only 2 trade partners — critical supply dependency"
    ]
  },
  "priceShock": {
    "score": 58,
    "volatileIndicators": 4,
    "supplyRiskFactors": 3,
    "shockLevel": "HIGH",
    "signals": [
      "Commodity prices trending up 15%+ — supply pressure building",
      "3 supply disruption factors — price shock catalyst"
    ]
  },
  "allSignals": [
    "4 severe weather alerts — crop-threatening conditions",
    "3 drought-condition days — water stress risk",
    "3 extreme temperature events — crop damage risk",
    "Trade HHI 2840 — concentrated commodity supply chain",
    "Only 2 trade partners — critical supply dependency",
    "Commodity prices trending up 15%+ — supply pressure building",
    "3 supply disruption factors — price shock catalyst"
  ],
  "recommendations": [
    "Severe weather stress — activate crop insurance and hedging strategies",
    "Concentrated trade dependency — diversify supplier base",
    "Price shock risk — consider commodity futures hedging",
    "Elevated commodity risk — increase inventory buffers and forward contracts"
  ]
}

Individual tool output: weather_yield_threat_monitor

{
  "region": "Mato Grosso, Brazil",
  "yieldThreat": {
    "score": 44,
    "level": "MODERATE_STRESS",
    "severeAlerts": 2,
    "signals": [
      "2 severe weather alerts — crop-threatening conditions",
      "2 agricultural-impact disasters — regional crop threat"
    ]
  },
  "disasters": [
    {
      "eventType": "flood",
      "alertLevel": "orange",
      "affectedCountries": ["Brazil"],
      "date": "2026-03-18"
    }
  ]
}

Individual tool output: trade_dependency_analysis

{
  "commodity": "soybeans",
  "tradeDisruption": {
    "score": 72,
    "tradePartners": 2,
    "concentrationHHI": 4120,
    "disruptionLevel": "CRITICAL_DEPENDENCY",
    "signals": [
      "Trade HHI 4120 — concentrated commodity supply chain",
      "Only 2 trade partners — critical supply dependency",
      "5 price volatile periods — commodity price shock risk"
    ]
  },
  "tradeData": [...],
  "priceData": [...]
}

How much does it cost to run agricultural commodity risk analysis?

This MCP server uses pay-per-event pricing — you pay $0.045 per tool call. All 8 tools, including compare_commodity_risks, are priced equally. Platform compute costs are included.

You can set a maximum spending limit per run to control costs. The server checks eventChargeLimitReached before each tool execution and returns a structured error if the budget is reached — it never silently over-charges.

The Apify Free plan includes $5 of monthly credits, enough for 111 tool calls at no cost. Compare this to specialized agricultural data platforms (Bloomberg Commodity: $2,000+/month; Gro Intelligence: $500–2,000/month) — most users spend $1–25/month with no subscription commitment.

Tips for best results

1. Use compare_commodity_risks for initial screening. It runs all 8 sources and returns all four scores in one call. Drill into individual tools only when you need targeted detail or faster response times.

2. Pass a specific growing region alongside the commodity. { "commodity": "corn", "region": "Iowa" } returns more targeted weather alerts than passing just the commodity name. Nominatim geocodes the region to coordinates used by downstream weather sources.

3. Set a spending limit for automated agent workflows. When scheduling recurring calls, set maxTotalChargeUsd to cap daily spend. The server returns a clean JSON error when the limit is reached so your agent can handle it gracefully.

4. Interpret HHI alongside partner count. An HHI of 2,800 with 3 partners is less fragile than an HHI of 2,800 with 2 partners. The signals array calls out the 2-partner threshold explicitly.

5. Combine with price_shock_probability before hedging decisions. The tool specifically compares the recent third of the FRED price series against the earlier third — a rising trend combined with active supply disruption signals (supplyRiskFactors ≥ 3) is the strongest composite hedge trigger the model produces.

6. Schedule daily growing season forecasts during critical phenology windows. For crops with narrow heat or frost tolerance windows (e.g., winter wheat vernalization, corn pollination), a daily growing_season_forecast call is more useful than weekly checks during those 2–4 week periods.

7. Pipe allSignals directly to your LLM context. The allSignals array in compare_commodity_risks is designed as a ready-to-use context block — feed it directly to your AI agent for narrative summary generation.

Combine with other Apify actors and MCP servers

Limitations

• NOAA weather alerts are US-centric. Global growing regions outside North America may return fewer NOAA alerts. GDACS disaster data provides the best global coverage for non-US regions, but has a higher minimum severity threshold than NOAA.
• GBIF pest data has observation lag. GBIF records depend on field reporting from researchers and citizen scientists. Outbreak detections will appear faster in official plant health databases than in GBIF occurrence records.
• FRED covers major commodity price indices, not spot prices. The price volatility model uses index series and CPI components, not exchange-quoted spot or futures prices. For futures market signals, integrate a specialized market data feed alongside this MCP.
• Trade data is not real-time. UN COMTRADE data is updated monthly to annually depending on the country and commodity. Trade concentration scores reflect historical flows, not breaking disruptions.
• World Bank agricultural indicators are annual. The agVulnerability component of the pest emergence model uses the latest available World Bank annual data, which may be 1–2 years behind current conditions.
• Weather data is not crop-stage-aware. The Weather Stress Index scores temperature extremes and precipitation deficits uniformly, without knowledge of the specific crop's growth stage. A -1°C frost has very different impacts on a germinating crop versus a dormant one — that phenological context must be provided by the calling agent.
• HHI calculation depends on COMTRADE record coverage. If COMTRADE returns sparse results for an uncommon commodity, the HHI may understate concentration. Review tradePartners count; fewer than 3 results warrants a manual verification.
• Response times vary by data source availability. Typical response is 30–90 seconds. GDACS and GBIF can occasionally take longer during high-traffic periods. The sub-actor timeout is set to 120 seconds per source.

Integrations

• Zapier — trigger commodity risk checks from Google Sheets commodity watchlists and push alerts to Slack or email
• Make — build automated morning risk briefing workflows that call multiple tools and format results for distribution
• Google Sheets — pipe daily weather stress scores and composite risk verdicts into a live commodity dashboard
• Apify API — integrate tool calls directly into trading system middleware or risk management platforms via REST
• Webhooks — fire alerts to a trading desk Slack channel when composite risk scores cross ELEVATED or HIGH_RISK thresholds
• LangChain / LlamaIndex — register this MCP server as a tool in your LangChain agent for natural language commodity risk queries in RAG pipelines

Troubleshooting

Composite score seems low despite known drought conditions. Check the severeAlerts and extremeTemps values in the weatherStress output. NOAA alert coverage for international regions is limited — use the region parameter with a specific country and province name to maximize geocoding accuracy. If the region is outside North America, the gdacs component carries more weight; confirm that GDACS has active records for the area.

trade_dependency_analysis returns concentrationHHI: 0 and tradePartners: 0. COMTRADE results depend on the commodity name matching trade description terminology. Try standard commodity names: "wheat", "soybeans", "crude palm oil", "Arabica coffee". Highly specific commodity names may not match COMTRADE HS code descriptions.

pest_emergence_alert returns CLEAR for a known outbreak region. GBIF observation lag means newly reported outbreaks may not appear for weeks. The recentObs count (observations from 2024 onward) is the leading indicator — a zero result means no records in GBIF, not confirmed absence. Cross-reference with FAO EMPRES-i or national plant health authority databases for outbreak confirmation.

Tool call times out. Sub-actors are allocated 120-second timeouts. If a specific data source is slow, the affected array will return empty and scoring will proceed on available data. Check which sub-arrays in the response are empty to identify the slow source. Re-running the call usually resolves transient timeouts.

Spending limit reached before expected. The maxTotalChargeUsd limit applies across all tool calls in a single run. If your agent is calling multiple tools in a session, the cumulative cost may reach the limit sooner than a single-tool calculation suggests. Increase the limit or structure your agent to batch compare_commodity_risks calls rather than calling individual tools for the same commodity.

Responsible use

• This MCP server accesses only publicly available data from government and international organization databases.
• All upstream data sources (NOAA, GDACS, UN COMTRADE, World Bank, GBIF, FRED) are provided under open data licenses.
• Risk scores are quantitative signals for informed decision-making, not certified financial or insurance advice.
• Do not use automated risk scores as the sole basis for material financial decisions without human review.
• For guidance on data use and web scraping legality, see Apify's guide.

❓ FAQ

How much does it cost to run agricultural commodity risk analysis with this MCP server? Every tool call — including compare_commodity_risks — costs $0.045. A full daily monitoring workflow covering 10 commodities costs $0.45. The Apify Free plan includes $5 of monthly credits (111 free calls). There is no subscription or minimum commitment.

What commodities can this MCP server analyze? Any commodity with UN COMTRADE trade records and a FRED price index: wheat, corn, soybeans, rice, coffee, cocoa, palm oil, sugar, cotton, canola, barley, sorghum, and more. Weather analysis covers any growing region that can be geocoded — from Iowa to the Cerrado to the Ukrainian steppe.

How does the Herfindahl-Hirschman Index (HHI) work in trade_dependency_analysis? The model extracts each trading partner's share of total commodity trade value from COMTRADE records, squares each share, and sums them (multiplied by 10,000). HHI below 1,500 indicates competitive, diversified supply. HHI above 2,500 indicates monopolistic concentration — the antitrust threshold used by the US Department of Justice. The signals array calls out HHI above 2,500 explicitly.

Is the weather data real-time? NOAA weather alerts and GDACS disaster events are near real-time (updated every 15–60 minutes). Weather forecast data provides multi-day outlooks. FRED price data and World Bank indicators are updated less frequently — monthly and annually respectively.

How is this different from Bloomberg's agricultural commodity data? Bloomberg provides exchange-quoted spot and futures prices with tick-level granularity. This MCP server provides risk intelligence from public government and international databases — NOAA, GDACS, World Bank, GBIF, COMTRADE, FRED — synthesized into scored risk dimensions. The two are complementary: this server is suited for AI agent integration and programmatic monitoring; Bloomberg is suited for direct trading execution. Cost is also radically different: $0.045/call vs. $2,000+/month.

Can I use this MCP server with an automated trading agent? Yes. The structured JSON output with verdict, compositeScore, and recommendations fields is designed for programmatic consumption. Your agent can check if verdict === 'CRITICAL' or 'HIGH_RISK' and trigger hedging workflows. Set a maxTotalChargeUsd spending limit to control costs in autonomous agent loops.

How long does a typical tool call take? 30–90 seconds for most tools. compare_commodity_risks runs all 8 data sources in parallel via Promise.all and typically completes in 45–90 seconds depending on source latency. The server runs in Apify Standby mode so there is no cold-start delay.

What pest species does pest_emergence_alert detect? The model scans GBIF scientific names for 10 invasive agricultural threat keywords: locust, armyworm, borer, weevil, aphid, rust, blight, fusarium, phytophthora, and whitefly. These cover the major global crop pathogens and invertebrate pests. The invasiveCount field tells you how many matching species were detected.

Can I schedule commodity risk monitoring to run automatically? Yes. Use Apify Schedules to trigger the MCP server on a recurring basis (daily, weekly, or custom cron). Combine with Apify Webhooks to send alerts to Slack, email, or a trading system endpoint when specific risk thresholds are crossed.

Is it legal to use this data for financial decision-making? All upstream data sources are publicly available open data from government agencies and international organizations. The data is free to use commercially under their respective open data licenses. Risk scores are analytical outputs, not licensed financial advice — see a qualified financial advisor for regulated investment decisions. For scraping legality guidance, see Apify's guide.

What happens when a data source returns empty results? Each sub-actor result defaults to an empty array on failure. Scoring models are designed to handle sparse data — a missing COMTRADE result produces tradePartners: 0 and a zero concentration score rather than an error. The composite score will be lower than with full data coverage, which may understate risk rather than overstate it. Check the raw data arrays in the response to confirm which sources returned results.

How is food_security_vulnerability different from compare_commodity_risks? food_security_vulnerability is country-focused: it assesses a nation's aggregate exposure to food supply shocks by querying agricultural GDP share, food import trade concentration, disaster history, and food price inflation. compare_commodity_risks is commodity-focused: it assesses a specific crop's market risk from weather, pests, trade, and price dimensions. Use the country tool for sovereign risk analysis and the commodity tool for supply chain and trading decisions.

Help us improve

If you encounter issues, you can help us debug faster by enabling run sharing in your Apify account:

1. Go to Account Settings > Privacy
2. Enable Share runs with public Actor creators

This lets us see your run details when something goes wrong, so we can fix issues faster. Your data is only visible to the actor developer, not publicly.

Support

Found a bug or have a feature request? Open an issue in the Issues tab on this actor's page. For custom commodity monitoring integrations or enterprise deployments, reach out through the Apify platform.