CyberMax-tools/recallroll-vehicle-recalls

Recallroll: US vehicle safety recalls, last 12 months

Need it fresh, filtered or via API? This free file is a snapshot (NHTSA recalls of the last 12 months), last updated 2026-09-25.

• US Vehicle Recall Brief: Weekly Edition ($9/month): a fresh NHTSA recall brief every Monday (PDF + 5 CSVs): every new campaign ranked, stop-use advisories first, 90-day trends and a script that checks your own vehicle list.
• Recallroll VIN + recall API (50 free calls a day, then $9/month for 2,000 calls): decodes up to 10 VINs per call and lists every NHTSA recall for each vehicle, Do Not Drive first, by API or MCP.
• All Three Monday Briefs: Rates, Recalls + Exploited Vulns ($29/month): all three weekly briefs (Treasury/BLS rates & macro, NHTSA recalls, CISA KEV/EPSS) for $29/month instead of $37.
• Get an email when this dataset updates: free, double opt-in, unsubscribe any time.

Every US safety recall campaign NHTSA received in the last 365 days (2025-09-25 to 2026-09-18): 958 campaigns with 43,450,237 vehicles and items potentially affected (manufacturer estimates), 50 of them in the last 30 days. 25 carry a Do Not Drive advisory and 24 a Park Outside (fire risk) advisory. Rebuilt from NHTSA's daily recall file; built 2026-09-25 01:29 UTC.

Free lookup page (decode a VIN, see its recalls, bulk-check 50 VINs, CSV): https://cybermax-tools-cybermax.static.hf.space/recallroll/index.html

Configs

Config	One row per	Rows	Use it for
recalls (default)	recall campaign	958	what was recalled, why, the risk, the fix, units affected, advisories
vehicles	campaign × make × model × model year	5,683	joining to your own inventory or fleet list by make/model/year

recalls columns

campaign (NHTSA recall number, e.g. 26V601000), report_received, owners_notified, manufacturer, type (vehicle / equipment / tire / child seat), component, units_affected, initiated_by (MFR = manufacturer, ODI/OVSC = NHTSA), mfr_campaign, do_not_drive, park_outside, summary, consequence, remedy, nhtsa_url, makes, vehicles (e.g. "FORD F-150 (2021-2023)"), vehicle_count.

vehicles columns

campaign, make, model, model_year (null when NHTSA lists no year), report_received, component, type, do_not_drive, park_outside.

Quick start

from datasets import load_dataset
recalls = load_dataset("CyberMax-tools/recallroll-vehicle-recalls", "recalls", split="train").to_pandas()
print(recalls[recalls.do_not_drive][["campaign", "vehicles", "component"]])

Or without downloading: https://datasets-server.huggingface.co/search?dataset=CyberMax-tools/recallroll-vehicle-recalls&config=vehicles&split=train&query=F-150

Limits

• A recall listed for a model year does not mean every vehicle of that year is affected (it may cover a build-date range or one plant), and it may already be repaired. Confirm a specific VIN at https://www.nhtsa.gov/recalls.
• Model names are NHTSA's recall spellings (e.g. "MUSTANG MACH E"), which can differ from VIN-decoder names.
• Information only: not safety, legal or purchase advice. Not affiliated with NHTSA or any manufacturer.

Source and licence

NHTSA recall flat file FLAT_RCL_POST_2010.zip (https://static.nhtsa.gov/odi/ffdd/rcl/), field definitions in RCL.txt there. U.S. Government work, no copyright in the U.S. Compiled by CyberMax.

Quickstart notebook

Load this dataset with pandas and see real outputs in the quickstart notebook.

Browse it on the web

Free, no sign-up: one page per recall campaign: https://data.cybermax-tools.workers.dev/recall/, plus feeds of new recalls (RSS, JSON). AI agents can query it through the remote MCP server https://data.cybermax-tools.workers.dev/mcp (no key).