Claude PopHIVE Public Health Data connectorWhat Claude can pull from PopHIVE datasets on your desktop.
The Claude PopHIVE Public Health Data connector exposes 5 tools: 5 read, 0 write. It is a desktop extension from an independent developer that lets Claude filter, compare and chart PopHIVE datasets inside Claude Desktop, with no account involved.
Verified Trustpilot reviews · AI, automation & growth agency
What this extension adds to Claude Desktop
PopHIVE is Yale's platform for near real-time US health data. This extension, published by the developer Cicatriiz rather than by Yale, puts several of its datasets within Claude's reach on your own computer: immunisation, respiratory surveillance, wastewater, chronic disease, hospital capacity, overdose deaths and youth mental health visits. You ask in plain words; Claude picks a dataset and explains what it finds.
Line states up against each other. compare_states sets one health metric for several states next to each other and, as the developer describes it, adds statistical analysis to the comparison.
Follow a curve over the years. time_series_analysis looks at how a measure changes over time, with options to aggregate the points, which suits series such as the national injury and overdose data for 2019-2025.
Start from the catalogue. get_available_datasets returns the catalogue of every dataset the extension reaches, before any narrowing down.
What it will not do: none of the 5 tools changes anything, and the extension only runs in Claude Desktop. Some datasets stop at national level, so a state filter on them comes back empty. Nothing runs by itself either: Claude queries only while it answers. Yale also runs its own remote PopHIVE connector, a separate entry in the directory. For scheduled reporting, look at automation platforms from our Integrations hub.
Five words before you install
Claude's vocabulary for this extension, in one minute.
- Connector
- The link you add once so Claude can draw on an outside source while it answers you. Here it takes the form of a desktop extension.
- Tool
- A single named capability exposed to Claude, such as filtering a dataset. Claude decides which ones to call; the directory names each.
- Approval
- The go-ahead Claude asks of you when a step would modify something in an account. Read-only lookups like these seldom trigger it.
- Directory
- The official catalogue where each connector or extension has a sheet with its publisher, purpose and tool list.
- MCP
- The shared standard behind connectors and extensions, the part that lets an assistant talk to an outside data source.
Install PopHIVE Public Health Data in three steps
- 01
Find it in Claude Desktop
From Claude Desktop only, head to Customize, then Connectors, and find the PopHIVE Public Health Data entry. In a Team or Enterprise organisation, it may first need switching on by an Owner or a Primary Owner.
- 02
Install the extension
The developer's README describes the route outside the directory too: download the extension file from the project's releases, open it with Claude Desktop and confirm in the install dialog. Either way, it ends up installed on your machine rather than in the cloud.
- 03
Set its options
According to the README, the install dialog then offers a few settings, such as how often the data refreshes and how large the local cache may grow. There is no sign-in screen: the data is public, so no account is linked.
The 5 tools, sorted by what they do
PopHIVE Public Health Data gives Claude 5 tools: 5 that read your account, 0 that change something in it.
All five read, and the developer's README gives each one a description. Each tool keeps its original name below.
- 5 read
What Claude reads (5)
5 toolsFive tools that query public datasets without changing them.
filter_data
Narrows a dataset down by state, date range, demographic group or condition, so Claude works on the slice you care about instead of the whole file.
compare_states
Sets one health metric for several states side by side, and the README says it comes with statistical analysis of the comparison. It works across states, not counties.
time_series_analysis
Analyses how a measure evolves over time; the README mentions aggregation options, so the points can be grouped rather than read one by one. It describes a trend, not a single snapshot.
get_available_datasets
Returns what the developer calls a comprehensive catalogue of the datasets available through the extension. It answers "what is in here" before any filter or comparison is run.
search_health_data
Searches across all datasets for a specific condition or keyword, rather than inside one dataset at a time. It is the entry point when you know the subject but not the dataset that holds it.
What Claude asks before acting
Claude's standing rule is to pause for your confirmation whenever it is about to act on an account in your name. With this extension that seldom applies: every documented tool reads public data.
In Team and Enterprise organisations, it is the owners who choose whether members can skip the prompt for certain actions. Owners can also narrow what an extension is allowed to do for everyone, and no member can lift that limit individually. Claude never goes beyond the rights of whoever linked a service; here there is no service account at all, only public datasets processed on your computer. Official per-tool read or write labels appear only inside a connected account's settings.
Which plans include it
Across the 819 entries of the official directory, not one lists the plans it runs on. For this extension, as for every connector, that answer is published nowhere.
What is published is the general rule: an extension like this one is installed in Claude Desktop, and every user can add desktop extensions there. In Team and Enterprise organisations, members wait until an Owner or a Primary Owner has switched it on. For its current state, check the extension's sheet in the official directory. Note that only its developer documents it; no Claude help article does.
Where this extension stops
An extension is not an automation. Claude calls these tools while it replies; nothing runs when a new data point lands.
A tool list showing only reading is an observed floor, not a guarantee. Being verified in the directory does not mean the code was audited for security; each sheet carries Anthropic's reminder that the publisher, not Anthropic, picks the tools and answers for them. This one comes from an individual developer, not from Yale, so read the project before installing. Its README flags the limits below.
The developer is also the only source on how it works.
Need help connecting PopHIVE Public Health Data to Claude?
A person reads every message.
