Hartford Auto-Injury Case Data
This site provides API access to data on car accident injury lawsuits in Connecticut's Hartford Judicial District. All data is derived from public court records and encompasses timelines, outcomes, and verdict and settlement-offer amounts broken down by injury type, accident type, and attorney. Please note that the data provided here is not an official record. Verify all case-specific information through the Court's public web portal.
You can query the API for free over HTTP or use the MCP connector to ask questions in plain language through an AI assistant (Claude, ChatGPT, Cursor).
Docs · Docs as markdown · llms.txt · OpenAPI · Terms · Privacy
Motivation
This is a side project I built to help a friend understand the cost-benefit of suing after a car accident, e.g. what his kind of case is typically worth, how long the process takes, how likely it was to get a settlement vs going to trial.
Court filings are free to access in CT, but you can only search one case at a time with no way to filter based on outcome or injury context. There are some vendors like LexisNexis and Trellis that provide access to high-level statistics, but these are paywalled at high prices and/or rely on attorney reporting and so skew towards outlier outcomes, and do not support filtering on things like injury type.
Here I've compiled every auto-injury case in the district and labeled each with relevant metadata pulled from the filings, so summary statistics reflect the entire corpus.
Connect
Claude
Add a custom connector with this URL (no authentication):
https://ctinjurycases.com/mcp
Claude Code:
claude mcp add --transport http ct-injury-cases https://ctinjurycases.com/mcp
ChatGPT
In developer mode, create a connector for this MCP server URL, with no authentication:
https://ctinjurycases.com/mcp
Cursor
In .cursor/mcp.json:
{ "mcpServers": { "ct-injury-cases": { "url": "https://ctinjurycases.com/mcp" } } }
curl
curl "https://ctinjurycases.com/v1/describe_dataset?format=md"
Example
What do juries award after a crash, by the kind of injury alleged? Net verdicts are after reductions for the plaintiff's share of fault. A case alleging several injuries counts under each, and injuries with fewer than 5 verdicts are left out here.
| Injury | Verdicts | Median net verdict | Middle half |
|---|---|---|---|
| fracture | 8 | $110,184 | $64,055 – $140,000 |
| laceration | 10 | $72,678 | $30,000 – $312,161 |
| concussion | 9 | $40,000 | $30,000 – $231,023 |
| disc herniation | 17 | $26,499 | $12,411 – $100,000 |
| other | 126 | $24,247 | $10,000 – $50,860 |
| sprain strain | 83 | $18,287 | $8,657 – $40,000 |
Run it on the live data: GET https://ctinjurycases.com/v1/stats?metric=net_verdict&group_by=injury_kind&format=md
Limits
Data is limited to Hartford, CT with a few simple supported access patterns, and endpoints are rate-limited to minimize hosting costs. There's a lot more that could be built, as there's a mountain of useful unstructured data in these documents, but I've focused only on my specific use case to start.
Things it would be great to do include expanding coverage statewide or to different states, extracting new types of data from filings (e.g. breaking down damages awarded by past/future, medical expenses, lost wages, pain and suffering, etc), more analytics options, a UI, etc.
My time is limited right now, but if you'd be interested in supporting the development of new features or partnering for higher-volume access, please get in touch below and let's chat!
Get in touch
Feedback, inquiries, ideas, issues.