Product Safety Consumer Risk MCP Server

Product safety intelligence for any product, manufacturer, or category — pulled from CPSC, NHTSA, FDA, and CFPB databases and scored through four analytical models. This MCP server gives compliance teams, insurance underwriters, and product risk analysts structured recall data, pre-recall warning signals, and a composite Product Risk Radar Score (0-100) in a single tool call.

The server runs up to 7 Apify data-collection actors in parallel, merges the results, and applies four scoring algorithms: Product Risk Radar, Pre-Recall Signal Detector, Supplier Risk Concentration Index, and Regulatory Enforcement Momentum. No scraping setup, no database subscriptions — connect via any MCP-compatible AI client and start querying immediately.

What data can you access?

Why use the Product Safety Consumer Risk MCP?

Tracking product safety manually means checking the CPSC recall portal, the NHTSA complaint database, the FDA recall enforcement page, and the CFPB complaint database separately — then reconciling results with no common scoring framework. That is four portals, four logins, no signal aggregation, and no early-warning logic.

This MCP collapses all four sources into a single structured query. More importantly, it goes beyond active recalls and surfaces pre-recall patterns: complaint clustering by issue type, pricing anomalies consistent with distressed inventory clearance, and website content changes that historically precede formal recall announcements.

• Standby mode — the server is always running, so MCP tool calls respond in seconds with no cold-start wait
• Parallel data collection — all 7 actors fire simultaneously, not sequentially, reducing response time
• API access — integrate into Python, JavaScript, or any HTTP client with a single POST request
• Spending limits — every tool call checks Actor.charge() and stops cleanly if your configured budget is reached
• MCP-native — works in Claude Desktop, Cursor, Windsurf, Cline, and any MCP-compatible client without custom code

Features

• Product Risk Radar Score (0-100) — a composite score weighted across 4 sub-models: Product Risk (30%), Pre-Recall Signals (25%), Supplier Risk (20%), Enforcement Momentum (25%)
• CPSC severity weighting — hazard descriptions are scanned for death, fire, electrocution (8 pts), burn, laceration, choking (5 pts), or general hazard (3 pts) before contributing to the recall score
• NHTSA consequence scoring — recall consequence fields are checked for crash, death, and fire keywords to weight automotive recall severity before capping at 25 points
• FDA Class I/II/III detection — medical device recalls are classified by severity class; Class I (life-threatening) scores 10 points per recall vs 5 for Class II, capped at 25
• Pre-recall complaint clustering — CFPB complaints are grouped by issue type; clusters of 3+ identical complaints score 7 points each plus 2 points per complaint count, identifying systemic problems before formal action
• Pricing anomaly detection — products with 30%+ price drops score 5 points each; 50%+ drops score 8 points each, detecting distressed-inventory clearance patterns that precede recall liquidation
• Complaint acceleration detection — compares last 30 days of CFPB complaints vs all prior history; a 2× acceleration triggers a pre-recall signal worth up to 15 points
• Website content removal signals — manufacturer page deletions, safety-URL changes, and major content modifications score up to 20 points in the pre-recall model
• Supplier concentration analysis — recall data is attributed by manufacturer/recalling_firm field; concentration above 50% by a single supplier triggers a SINGLE_SOURCE warning
• Repeat-offender detection — any manufacturer with 2+ recalls across CPSC/NHTSA/FDA is flagged; each repeat offender adds 8 points to supplier risk
• Enforcement recency scoring — recalls within the last 90 days score 7 points each in the enforcement momentum model, distinguishing active enforcement from historical records
• Cross-agency coordination detection — enforcement activity across 2 or 3 agencies simultaneously scores 15-25 extra points, flagging coordinated regulatory actions
• 7 MCP tools, pay-per-query — call exactly the tool you need; individual tools cost $0.045, the comprehensive report costs $0.045 (billed via Apify's pay-per-event system)

Use cases for product safety intelligence

Product liability insurance underwriting

Insurance underwriters pricing product liability policies need to score recall probability before binding coverage. Manually reviewing CPSC and NHTSA databases for a single product line takes hours and produces no defensible score. The generate_product_safety_report tool returns a 0-100 composite score, risk rating (SAFE through DANGER), and a breakdown by hazard type in seconds — giving actuaries a structured signal to incorporate into premium models.

Retail inventory risk management

Retailers and distributors stocking products from third-party manufacturers carry recall exposure on their warehouse inventory. The detect_pre_recall_signals tool identifies complaint clustering and pricing anomaly patterns before formal recall announcements arrive — giving procurement teams a 2-6 week window to reduce exposure or initiate supplier conversations while competitors are still unaware.

E-commerce marketplace compliance

Marketplace operators monitoring third-party seller listings for recalled or at-risk products can use search_active_recalls and monitor_pricing_anomalies together. Sudden price drops of 50%+ on specific product categories, combined with active recall records, identify sellers liquidating recalled stock. The Shopify Store Intelligence source surfaces low-rated storefronts that may be distributing products with known safety issues.

Product due diligence in M&A

Acquiring a consumer goods company requires understanding its product safety liability exposure before close. The assess_supplier_risk tool attributes all historical recalls by manufacturer name, identifies repeat offenders in the supply chain, and scores supplier concentration risk. Running track_enforcement_momentum across the target company's product categories establishes whether regulatory scrutiny is accelerating or declining.

Consumer safety compliance monitoring

Compliance officers at consumer goods manufacturers need ongoing monitoring of their own and competitors' product categories for recall trends. Scheduling periodic calls to track_enforcement_momentum detects complaint trend direction (RISING, STABLE, DECLINING) and cross-agency coordination signals that predict where enforcement attention is moving before formal announcements appear.

Regulatory affairs and recall response planning

Regulatory affairs teams at medical device and automotive companies benefit from knowing when complaint velocity is accelerating before the FDA or NHTSA initiates a formal inquiry. The pre-recall scoring model — complaint acceleration plus cluster detection — identifies the statistical signatures that historically appear 2-6 months before formal recall initiation, enabling proactive internal review.

How to use product safety intelligence via MCP

1. Connect the MCP server — add the server URL to your MCP client configuration. In Claude Desktop, paste https://product-safety-consumer-risk-mcp.apify.actor/mcp under mcpServers in claude_desktop_config.json. The server is always running in standby mode — no start time.
2. Choose the right tool — use search_active_recalls for a quick recall check, detect_pre_recall_signals for early-warning analysis, or generate_product_safety_report for a full cross-source assessment with scoring and action recommendations.
3. Provide product or manufacturer context — all tools accept a product string (name, brand, or category). The assess_supplier_risk tool takes a manufacturer name. Optionally supply a url to detect_pre_recall_signals for targeted website change monitoring.
4. Read the structured JSON response — every tool returns scored results with a signals array describing what was found, a numeric score, a categorical risk level, and (for the full report) immediateActions and monitoringPriority arrays.

MCP tools

Tool input parameters

Input examples

Quick recall check for a product:

{
  "tool": "search_active_recalls",
  "arguments": {
    "product": "infant car seat",
    "manufacturer": "Graco"
  }
}

Pre-recall early warning with URL monitoring:

{
  "tool": "detect_pre_recall_signals",
  "arguments": {
    "product": "BabyBright baby monitor",
    "url": "https://www.babybright.com/products/monitor-pro"
  }
}

Full comprehensive safety report:

{
  "tool": "generate_product_safety_report",
  "arguments": {
    "product": "lithium battery charger",
    "manufacturer": "PowerCell Industries",
    "url": "https://www.powercell.com/safety"
  }
}

Input tips

• Be specific with product names — "Graco infant car seat" returns more focused recall results than "car seat" because the query is passed directly to CPSC, NHTSA, and FDA search
• Use generate_product_safety_report for high-stakes decisions — it runs all 7 actors simultaneously and produces the composite score; individual tools are better for fast spot-checks
• Supply the url parameter when available — website change monitoring is far more precise when given an exact product or safety page URL vs a keyword search
• Use sector in track_enforcement_momentum — qualifying with "automotive", "consumer", or "medical" focuses recall data on the relevant agency and reduces noise in mixed-category searches
• Start with search_active_recalls — if the score is below 20 (CLEAR), no further investigation is needed; escalate to generate_product_safety_report only when the initial score warrants it

Output example

The generate_product_safety_report tool returns a complete structured report. Example output for a hypothetical infant sleep product:

{
  "product": "DreamSafe infant bassinet",
  "compositeScore": 68,
  "riskRating": "WARNING",
  "productRisk": {
    "score": 71,
    "activeRecalls": 5,
    "vehicleRecalls": 0,
    "deviceRecalls": 0,
    "complaintVolume": 23,
    "riskLevel": "HIGH",
    "signals": [
      "4 CPSC recalls — active product safety concerns",
      "23 CFPB consumer complaints — elevated consumer dissatisfaction"
    ]
  },
  "preRecallSignals": {
    "score": 58,
    "complaintClusters": 3,
    "pricingAnomalies": 4,
    "contentChanges": 2,
    "signalStrength": "STRONG",
    "signals": [
      "3 complaint clusters detected — repeated issue patterns",
      "Cluster: \"infant suffocation hazard\" — 7 similar complaints",
      "Cluster: \"mesh wall collapse\" — 5 similar complaints",
      "4 significant price reductions detected — pre-recall liquidation pattern",
      "2 products with 50%+ price drops — possible distressed inventory clearance"
    ]
  },
  "supplierRisk": {
    "score": 44,
    "productCount": 214,
    "recallConcentration": 80,
    "shopifyPresence": true,
    "riskLevel": "CONCENTRATED",
    "signals": [
      "DreamSafe LLC accounts for 80% of recalls — concentrated supplier risk",
      "Recalls concentrated in 1-2 suppliers — single-source vulnerability"
    ]
  },
  "enforcementMomentum": {
    "score": 52,
    "recentRecalls": 3,
    "recallAcceleration": true,
    "complaintTrend": "RISING",
    "momentumLevel": "ACTIVE",
    "signals": [
      "3 recalls in last 90 days — active enforcement period",
      "Recall acceleration detected — enforcement pace increasing",
      "Complaint trend RISING: 14 last 30 days vs 9 prior 30 — enforcement likely to follow"
    ]
  },
  "allSignals": [
    "4 CPSC recalls — active product safety concerns",
    "23 CFPB consumer complaints — elevated consumer dissatisfaction",
    "3 complaint clusters detected — repeated issue patterns",
    "Cluster: \"infant suffocation hazard\" — 7 similar complaints",
    "4 significant price reductions detected — pre-recall liquidation pattern",
    "DreamSafe LLC accounts for 80% of recalls — concentrated supplier risk",
    "3 recalls in last 90 days — active enforcement period",
    "Complaint trend RISING: 14 last 30 days vs 9 prior 30 — enforcement likely to follow"
  ],
  "immediateActions": [
    "Pre-recall signals detected — prepare recall response plan"
  ],
  "monitoringPriority": [
    "Daily complaint monitoring for new clusters",
    "Weekly recall database checks across CPSC, NHTSA, FDA",
    "Supplier quality audit scheduling",
    "Regulatory calendar tracking for upcoming enforcement actions"
  ]
}

Output fields

How much does it cost to run product safety checks?

This MCP uses pay-per-event pricing — you pay $0.045 per tool call. The comprehensive report (generate_product_safety_report) is also $0.045. Platform compute costs are included.

You can set a maximum spending limit per run or per month in your Apify account to control costs. The actor stops cleanly when your budget is reached.

Compare this to commercial product safety databases that charge $2,000-15,000 per year for recall monitoring subscriptions. Most teams conducting 100-200 safety checks per month spend under $10 with this MCP — with no subscription commitment.

The Apify Free plan includes $5 of monthly credits, which covers approximately 111 tool calls — enough to run complete safety reports on over 100 products per month at no cost.

How the Product Safety Consumer Risk MCP works

Data collection via parallel actor orchestration

Each tool call invokes runActorsParallel, which fires multiple Apify actors simultaneously using Promise.allSettled. All sources — CPSC, NHTSA, FDA, CFPB, e-commerce pricing, Shopify, and website change monitor — run concurrently. Promise.allSettled (not Promise.all) ensures partial results are always returned even if one upstream source fails. Memory per actor run is capped at 256 MB with a 120-second timeout. Results are collected into a keyed Record<string, unknown[]> and passed to the scoring layer.

Four-model scoring layer

Product Risk Radar (scoreProductRiskRadar) — CPSC hazard keywords (death, fire, electrocution = 8 pts; burn, laceration, choking = 5 pts; other = 3 pts) feed a capped 25-pt sub-score. NHTSA uses crash/death/fire consequence detection similarly. FDA uses Class I = 10 pts, Class II = 5 pts, other = 2 pts, capped at 25. CFPB volume contributes 2 pts per complaint, capped at 25.

Pre-Recall Signal Detector (detectPreRecallSignals) — groups CFPB complaints by issue/sub_product field (first 50 chars); clusters of 3+ score 7 pts each plus 2 pts per complaint. Price fields (current_price/salePrice vs original_price/regularPrice) are normalized; discounts ≥30% score 5 pts, ≥50% score 8 pts. Website removals and safety-URL changes score up to 20 pts. A 2× complaint acceleration in the last 30 days triggers up to 15 additional points.

Supplier Risk Concentration Index (assessSupplierRisk) — manufacturer identity is normalized across manufacturer, company, firm_name, and recalling_firm fields. Top-manufacturer concentration ratio scores 0-40 pts. Each repeat offender (≥2 recalls across any agency) adds 8 pts, capped at 25. Shopify stores rated below 3.0 add 5 pts each; stores with 100+ products add 5 pts.

Enforcement Momentum Tracker (trackEnforcementMomentum) — recalls are binned into last-90-days vs last-12-months buckets using normalized date fields. Recency scores 7 pts per recent recall, capped at 35. Cross-agency activity scores 25 pts (3 agencies), 15 pts (2), or 5 pts (1). Complaint 30-vs-60-day bucket comparison sets RISING/STABLE/DECLINING trend (25/10/3 pts).

Composite report assembly

generateProductSafetyReport applies the weighted composite: Product Risk × 0.30 + Pre-Recall × 0.25 + Supplier × 0.20 + Enforcement × 0.25. Thresholds: SAFE (<20), WATCH (20-39), CAUTION (40-59), WARNING (60-79), DANGER (≥80). Immediate actions and monitoring priorities are generated programmatically from sub-score thresholds, not static templates.

Tips for best results

1. Use manufacturer name for supplier risk checks. The assess_supplier_risk tool works best with the exact manufacturer name as it appears in recall databases (e.g., "Graco Children's Products" rather than "Graco"). CPSC and FDA use legal entity names in the recalling_firm field.

2. Supply a URL for the most accurate pre-recall detection. When you pass a product page or safety page URL to detect_pre_recall_signals or generate_product_safety_report, the website change monitor targets that exact page. Without a URL, it runs a keyword search which is less precise for spotting specific page removals.

3. Combine search_active_recalls with monitor_pricing_anomalies for marketplace monitoring. Active recall records combined with 50%+ discount pricing on the same product is a strong indicator of recalled inventory being liquidated through secondary channels — pattern that neither tool alone can confirm.

4. Schedule track_enforcement_momentum weekly, not daily. Enforcement trends are meaningful at weekly or monthly intervals. Daily calls add cost without improving signal quality because agency databases typically update at lower frequency.

5. Escalate only on CAUTION (40+) scores. Start every product evaluation with search_active_recalls. Reserve generate_product_safety_report for products that score 40 or above — this keeps costs low and focuses full-depth analysis on products that warrant it.

6. Use sector to narrow vehicle vs consumer vs medical. Many product categories span multiple agencies. Adding sector: "automotive" to track_enforcement_momentum limits the recall trend analysis to NHTSA data, reducing cross-contamination from unrelated CPSC or FDA actions on similarly named product categories.

Combine with other Apify actors

Limitations

• CFPB covers financial products, not all consumer goods. The CFPB complaint database is specific to financial products (credit cards, mortgages, bank accounts). For non-financial consumer goods, the complaint acceleration signal relies on CPSC hazard reports rather than CFPB volume, which limits pre-recall detection for pure consumer product categories.
• Date field normalization depends on upstream data consistency. Recall recency scoring parses date, recall_date, report_date, and event_date fields. If the upstream actor returns dates in an unparseable format, those records are excluded from time-based calculations. The total recall count remains accurate; only the recency sub-score is affected.
• Pricing anomaly detection requires e-commerce listings to be present. Products sold exclusively through direct channels, B2B, or without consistent list-price data will show zero pricing anomaly score regardless of actual recall risk.
• Website change monitoring is a snapshot, not continuous tracking. The detect_pre_recall_signals tool takes a point-in-time snapshot. For true continuous monitoring, schedule recurring calls or use the Website Change Monitor actor directly with Apify Scheduler.
• Manufacturer name normalization is heuristic. Supplier concentration analysis normalizes manufacturer names by checking four possible fields. Companies that file recalls under subsidiary names that differ from their brand name may appear as separate manufacturers, understating concentration.
• No food, drug, or chemical recalls. This MCP covers CPSC (consumer products), NHTSA (vehicles), FDA medical devices (not drugs), and CFPB. FDA food recalls and FDA drug recalls are handled by separate actors. Chemical and pesticide recalls (EPA) are not currently included.
• Historical data depth depends on upstream actors. Each agency actor returns recent records; the depth of historical data varies by agency. Enforcement momentum calculations are most reliable when agencies have returned at least 5 records with parseable dates.
• No real-time push alerts. Scores are computed on demand per tool call. For threshold-based alerting, use Apify Webhooks to trigger downstream notifications when a scheduled run's output exceeds a score threshold.

Integrations

• Apify API — POST directly to the MCP endpoint from any backend service; include Authorization: Bearer YOUR_APIFY_TOKEN and the MCP JSON-RPC body
• Webhooks — trigger Slack, email, or webhook notifications when a scheduled product safety run produces a score above a defined threshold
• Zapier — build no-code workflows that run product safety checks on new inventory additions and push results into Airtable, Notion, or Google Sheets
• Make — schedule periodic product safety scans across your supplier list and route WARNING/DANGER-rated products to a review queue automatically
• Google Sheets — export scored product safety reports to a tracking spreadsheet for compliance documentation and audit trails
• LangChain / LlamaIndex — use this MCP as a tool within agentic safety research workflows that chain recall lookup with policy retrieval and risk memo drafting

How to connect this MCP server

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "product-safety-consumer-risk": {
      "url": "https://product-safety-consumer-risk-mcp.apify.actor/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_APIFY_TOKEN"
      }
    }
  }
}

Python (via HTTP)

import requests

url = "https://product-safety-consumer-risk-mcp.apify.actor/mcp"
headers = {
    "Content-Type": "application/json",
    "Authorization": "Bearer YOUR_APIFY_TOKEN"
}

payload = {
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
        "name": "generate_product_safety_report",
        "arguments": {
            "product": "lithium battery charger",
            "manufacturer": "PowerCell Industries"
        }
    },
    "id": 1
}

response = requests.post(url, json=payload, headers=headers)
data = response.json()
report = data["result"]["content"][0]["text"]
print(f"Composite score: {report}")

JavaScript

const response = await fetch(
  "https://product-safety-consumer-risk-mcp.apify.actor/mcp",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "Authorization": "Bearer YOUR_APIFY_TOKEN"
    },
    body: JSON.stringify({
      jsonrpc: "2.0",
      method: "tools/call",
      params: {
        name: "search_active_recalls",
        arguments: {
          product: "infant car seat",
          manufacturer: "Graco"
        }
      },
      id: 1
    })
  }
);

const data = await response.json();
const result = JSON.parse(data.result.content[0].text);
console.log(`Risk level: ${result.productRisk.riskLevel}, Score: ${result.productRisk.score}`);

cURL

# Call the search_active_recalls tool
curl -X POST "https://product-safety-consumer-risk-mcp.apify.actor/mcp" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "search_active_recalls",
      "arguments": {
        "product": "infant car seat",
        "manufacturer": "Graco"
      }
    },
    "id": 1
  }'

# Call the full safety report tool
curl -X POST "https://product-safety-consumer-risk-mcp.apify.actor/mcp" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_APIFY_TOKEN" \
  -d '{
    "jsonrpc": "2.0",
    "method": "tools/call",
    "params": {
      "name": "generate_product_safety_report",
      "arguments": {
        "product": "lithium battery charger",
        "manufacturer": "PowerCell Industries",
        "url": "https://www.powercell.com/safety"
      }
    },
    "id": 2
  }'

Troubleshooting

• Score is 0 despite known recalls — the query may be too broad or the product name differs from how the manufacturer appears in recall databases. Try the exact legal manufacturer name (e.g., "Graco Children's Products Inc" instead of "Graco"). Also check whether the product is classified under CPSC (consumer goods), NHTSA (vehicles), or FDA (medical devices) — using search_active_recalls queries all three simultaneously, but if the product type is niche, one agency may dominate.

• signalStrength is NONE despite known complaint problems — pre-recall signal detection depends on CFPB data, which covers financial products. If your product is a physical consumer good with complaints filed with CPSC rather than CFPB, the complaint clustering sub-score will be low. Run search_active_recalls to see the CPSC recall count directly — the productRisk.riskLevel field captures non-financial product risk more accurately for physical goods.

• detect_pre_recall_signals returns low pricing anomaly scores — pricing anomaly detection requires e-commerce listings with both a current price and an original/list price. Products without visible list prices (sold at a single price point) or sold exclusively on platforms that don't expose original price fields will score 0 on pricing anomalies regardless of actual market conditions.

• MCP tool call times out — the generate_product_safety_report tool fires 7 actors simultaneously; in rare cases upstream actors can take longer than the 120-second per-actor timeout. If this occurs, use individual tools (search_active_recalls, detect_pre_recall_signals) separately — each calls 3-4 actors instead of 7, completing faster.

• Spending limit reached message returned — this means your configured per-run spending cap was hit. The tool returns a structured error object ({ "error": true, "message": "Spending limit reached..." }) so your code can handle it gracefully. Increase the run spending limit in your Apify account settings or reduce call frequency.

Responsible use

• This MCP only accesses publicly available government databases (CPSC, NHTSA, FDA, CFPB) and publicly accessible e-commerce and website data.
• Recall and complaint data is factual public record. Use it accurately — do not represent risk scores as regulatory determinations.
• Comply with applicable data protection and competition laws when using scoring outputs to make decisions about suppliers, manufacturers, or products.
• Do not use pre-recall signal outputs to short-sell securities or trade on material non-public information derived from early warning signals.
• For guidance on web scraping legality, see Apify's guide.

FAQ

How early can product safety recall signals be detected before a formal announcement? The pre-recall signal model identifies complaint clustering and pricing anomaly patterns that historically precede formal recall announcements by 2-6 months. Complaint acceleration (a 2× spike in the last 30 days vs prior history) and issue clusters of 3+ identical complaints are the strongest early indicators. These are statistical patterns, not certainties — elevated scores warrant further investigation, not automatic recall conclusions.

What product categories does this MCP cover? The MCP covers consumer products regulated by CPSC (household goods, toys, electronics, furniture), vehicles and automotive components regulated by NHTSA, medical devices regulated by FDA, and financial products in the CFPB database. Food and drug recalls are not included — use the Food Safety Supply Chain MCP for food products and the FDA Drug Approvals MCP for pharmaceutical monitoring.

How accurate is the Product Risk Radar Score? The score aggregates objective data points (recall counts, complaint volumes, price changes, website modifications) through a weighted model. Accuracy depends on the quality and completeness of upstream agency data. The scoring is transparent — every point contribution is documented in the signals array so users can inspect what drove the score rather than treating it as a black box.

Can I monitor a product or manufacturer continuously over time? Yes. Schedule periodic calls to track_enforcement_momentum or generate_product_safety_report using Apify Scheduler or a cron-triggered API call. Comparing scores week-over-week reveals trend direction better than any single snapshot. Combine with Apify Webhooks to push alerts when a score crosses your defined threshold.

How is this different from commercial recall monitoring services? Commercial services like Recall.gov alerts, ECRM, or industry-specific monitoring platforms typically provide raw recall notifications without composite scoring, pre-recall signal detection, or supplier concentration analysis. This MCP adds four analytical layers on top of the same public data sources those services use. It is also pay-per-query with no subscription, making it more cost-effective for teams that need on-demand checks rather than continuous monitoring of a fixed product set.

Does the product safety report cover international recalls? The current data sources are US government databases: CPSC (US consumer products), NHTSA (US vehicles), FDA (US medical devices), CFPB (US financial products). International recall databases (EU RAPEX, UK OPSS, Health Canada) are not currently integrated.

Is it legal to use this data for insurance underwriting or investment decisions? All underlying data sources are US government public databases designed for public access. Using this data for risk assessment, underwriting, and investment research is consistent with the agencies' intent in publishing the data. Always consult legal counsel before making binding decisions based on automated risk scores. See Apify's guide on web scraping legality.

What happens if one of the 7 data sources is unavailable during a run? The actor client uses Promise.allSettled rather than Promise.all, so a single failing source returns an empty array rather than failing the entire request. The scoring models handle empty arrays gracefully — a missing source contributes zero to its sub-score. The response will still include results from all available sources, and the signals array will reflect only what was actually found.

Can I use this MCP with Cursor, Windsurf, or Cline? Yes. Any MCP-compatible client that supports HTTP-based MCP servers can connect to this endpoint. Add the server URL (https://product-safety-consumer-risk-mcp.apify.actor/mcp) with your Apify token as a Bearer authorization header in the client's MCP configuration.

How many product safety checks can I run per month on the free plan? Apify's free plan includes $5 of monthly credits. At $0.045 per tool call, that covers approximately 111 tool calls — enough to run complete generate_product_safety_report checks on over 100 products per month at no cost.

What is the difference between riskLevel and riskRating? riskLevel appears on each sub-model result (e.g., productRisk.riskLevel: "HIGH") and uses model-specific labels (CLEAR / LOW / MODERATE / HIGH / CRITICAL for product risk; DIVERSIFIED through CRITICAL for supplier risk). riskRating is the top-level composite rating (SAFE / WATCH / CAUTION / WARNING / DANGER) derived from the weighted composite score. Use riskRating for go/no-go decisions; drill into riskLevel fields to understand which dimension is driving the risk.

Can this MCP detect risks for products that have not yet been recalled? This is its primary purpose. The Pre-Recall Signal Detector is specifically designed to identify statistical patterns — complaint clustering, pricing anomalies, website content changes — that appear before formal recall announcements. A signalStrength of STRONG or IMMINENT does not confirm a future recall but indicates the product matches historical pre-recall patterns and warrants closer scrutiny.

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 solutions or enterprise integrations, reach out through the Apify platform.