Install the official Alpha Sophia plugin in Cursor — the AI code editor — and let the Agent query, rank, and size markets on real, current, sourced US provider data without leaving your editor. The plugin bundles our MCP server and a built-in playbook, so questions become target lists, enrichment scripts, and dashboards grounded in licensed claims data, not hallucinated NPIs.
Official Alpha Sophia plugin for Cursor
→ resolve_entity("plaque psoriasis", "dermatology")
→ search_hcp(... order_by: patient_volume)
→ wrote targets.csv (842 rows)
842 California dermatologists managing plaque psoriasis. Top rows by patient volume:
Pulled from live, sourced data — ready to commit or import.
Illustrative example of an Alpha Sophia–powered Agent run in Cursor.
Prefer Claude or ChatGPT? Get the Claude plugin or the ChatGPT connector, or see all the ways to connect AI agents to Alpha Sophia over MCP.
Cursor is where your team already writes scripts, queries, and tools. Connecting it to Alpha Sophia means the data is real where the work actually happens — not copy-pasted in from a separate tab.
Pull provider counts, lists, and market sizes directly into the file, notebook, or query you're working on — no exporting, no tab-switching.
A model left to itself will invent provider names and numbers. The MCP server makes the Agent retrieve them instead, so what lands in your script is real.
Turn a one-off question into a script you commit and re-run. Workflows live in your repo, reviewed and rerunnable, not lost in a chat.
The Agent can join Alpha Sophia data with your CSVs, SQL, and CRM exports already open in the workspace — enrichment without a pipeline project.
Trials, practice moves, and volumes shift every quarter. The Agent reads live data on each run, so your tooling never ships last year's numbers.
Every figure traces back to a sourced record — the audit trail HIPAA, the Sunshine Act, and MLR review demand, baked into how the data arrives.
Same task in your editor. The MCP server is the difference between placeholder data you'd have to replace and a real dataset you can ship.
"Write a script that outputs the top orthopedic surgeons for total knee replacement near Boston."
Writes clean code — then fills it with a // TODO: real data placeholder or a handful of invented names, because it has no live
provider data and a training cutoff.
The Agent calls Alpha Sophia, ranks real surgeons by procedure volume near Boston, and writes a populated, affiliation-linked CSV — code and data, done together.
"Match this conference attendee list to NPIs and add specialty and affiliation."
Can scaffold the matching logic, but has nothing authoritative to match against — so it guesses at NPIs or leaves the columns blank.
The Agent resolves each name against real records, fills in canonical NPIs, specialty, and primary affiliation, and flags the rows it couldn't confidently match.
Illustrative examples based on Alpha Sophia data.
The plugin gives the Cursor Agent both the tools (our MCP server) and the playbook (skills and rules), so it queries Alpha Sophia the same disciplined way our in-app assistant does — and writes the result into your workspace.
Install the Alpha Sophia plugin from the Cursor Marketplace (or run
/add-plugin),
then authenticate to your Alpha Sophia organization. The tools and
playbook now appear in Agent mode for everyone on your team who
installs it. Prefer just the data tools? Connect the MCP server
directly instead.
Ask in plain English — "how many", "find and rank", "size the market" — or go further and tell the Agent to write a script, populate a dataframe, or enrich the file you have open. No filter language to learn.
Behind the scenes, the Agent resolves your terms (a specialty, a drug, a procedure, a city) to canonical IDs, builds a structured filter, and runs it against Alpha Sophia — the same parse, build, execute pipeline as the app. Numbers come from data, not memory.
A populated CSV, a dataframe, a dashboard query, a small tool — with the filters the Agent applied stated in plain English so you can trust and refine it. Commit it, re-run it, or take the list into Alpha Sophia to export or sync to your CRM.
The same data backbone serves every team — so technical and commercial folks alike move from days of wrangling to minutes of building.
For: Enrichment & pipelines
For: CRM-ready target lists
For: Live data for dashboards
For: KOLs from evidence
For: Internal tools, no data team
For: Sites of care by geography
A single answer is useful. A finished, committed artifact is better. In Cursor's Agent mode, the Agent chains multiple Alpha Sophia calls and its own code into a complete piece of work.
A quick chat is plenty for a one-off count. For a populated, re-runnable artifact, let the Agent build it.
"Write and save a script that builds our ranked GI target list from live data and outputs a CRM-ready CSV."
"Build a KOL dataset for our indication — top authors and investigators with affiliations and recent output — as JSON."
"Create a notebook that sizes our therapy market by state with sites of care and procedure volumes, charts included."
Every Alpha Sophia entity is available to the plugin — the same data that powers the web app, scoped to your organization.
One install from the Cursor Marketplace, then authenticate. Your organization needs an Alpha Sophia subscription with the API and Agent modules enabled.
# 1. Open the marketplace inside Cursor (Customize → Plugins)
# or run the command:
/add-plugin
# 2. Search "Alpha Sophia", install, then restart Cursor
# 3. Sign in to your org the first time the Agent calls a tool
Browse to the Alpha Sophia plugin in Customize → Plugins (or run
/add-plugin), at the user or project
level.
On the first tool call, Cursor prompts you to sign in. Every answer is scoped to your licensed data and agent tokens.
The plugin ships our MCP server and the playbook together, using the same pipeline as the Alpha Sophia web app.
Add the Alpha Sophia MCP server under Settings → Tools & MCP, or
drop this into .cursor/mcp.json — you'll
get the tools without the bundled playbook.
{
"mcpServers": {
"alpha-sophia": {
"url": "https://api.alphasophia.com/v1/mcp"
}
}
}Want us to set it up with your team? Talk to our team.
Same data backbone — whether you're shipping a script or sizing a market.
Enrichment jobs and pipelines fed by real NPIs, specialties, and volumes — no scraping, no guesswork.
CRM-ready target lists and account scoring generated and re-run from the editor, not requested from an analyst.
Live counts and market-size figures wired straight into dashboards, models, and notebooks.
KOL datasets built from current publications and active trials, exportable and reproducible.
An analyst-grade data layer for your prototypes and internal tools — without staffing a data team.
Sites-of-care coverage and procedure volumes pulled on demand to back pricing and access decisions.
Access is authenticated per user and scoped to your organization, under the same enterprise-grade controls trusted by 100+ life science customers. Because answers are sourced from Alpha Sophia rather than the model's memory, the data in your scripts and tools is traceable and auditable. As with any AI in a regulated setting, keep a human in the loop before outputs inform irreversible actions or regulated content.
It's the official Alpha Sophia plugin for Cursor — the AI code editor. Like a Cursor plugin should, it bundles everything the Agent needs into one install: our MCP (Model Context Protocol) server plus a built-in playbook (skills and rules) so Cursor queries Alpha Sophia the same disciplined way our in-app assistant does. Once installed, Cursor's Agent pulls real, current, sourced US healthcare provider data straight into your editor, so the lists, counts, and scripts you build are grounded in licensed claims data instead of a model's best guess.
It's a full plugin. The Cursor Marketplace lets plugins package MCP servers together with skills, rules, and commands, and ours does exactly that — so the Agent doesn't just get the tools, it gets the expertise for using them correctly (how to read specialties, procedures, diagnoses, and how to resolve terms before filtering). If you'd rather wire up only the raw data tools, you can also connect the Alpha Sophia MCP server on its own — but the full plugin is the recommended, batteries-included path, just like our Claude plugin.
Same plugin idea and same data backbone, different home. The Claude plugin lives in Claude and Claude Cowork; the Cursor plugin lives in your IDE next to your repo, CSVs, SQL, and notebooks. Cursor is built for people who build — data and analytics engineers, RevOps, BI teams, and technical founders — so the Cursor plugin shines when you want to turn a question into a reusable script, a pipeline, or an internal tool, not just a one-off answer. Use whichever fits where your team works; the underlying Alpha Sophia tools and expertise are the same.
MCP (Model Context Protocol) is the open standard Cursor uses to talk to external tools and data, and it's the engine inside the plugin. The Alpha Sophia MCP server exposes tools like resolve_entity, plus search / count / market_size for providers and sites of care, and search / count for clinical trials, investigators, publications, and authors. When you ask Cursor's Agent a question, it resolves your terms to canonical IDs, builds a structured filter, and runs it against Alpha Sophia's Streamable HTTP endpoint — no raw SQL to write, no invented NPIs.
Open the marketplace inside Cursor (Customize → Plugins, or run /add-plugin), search for "Alpha Sophia", and install it at the user or project level — then restart Cursor and sign in to your Alpha Sophia organization the first time the Agent calls a tool. Prefer the data tools only? Add the MCP server directly under Settings → Tools & MCP (or in .cursor/mcp.json) with the URL https://api.alphasophia.com/v1/mcp.
No. Cursor's Agent answers plain-English questions the same way a chat assistant does — "how many", "find and rank", "how big is the market" — so commercial, marketing, and medical teams who already work in Cursor can use it without writing code. The difference is that when you do want a script, a notebook cell, or a repeatable workflow, the plugin drops the real data straight into what you're building.
Anything that needs real provider data in your workflow: target-list scripts and ranked shortlists, market-sizing models, data-enrichment jobs that match a CSV to NPIs, BI dashboards fed from live counts, KOL maps from publications and trials, and small internal tools — all grounded in Alpha Sophia and version-controlled in your repo. Cursor handles the building; the plugin supplies the trustworthy data and the playbook for using it well.
Yes — Agent mode is where it's best. The Agent can chain several Alpha Sophia calls together (resolve a condition and geography, pull and rank providers, cross-reference KOLs, then write the result to a file or dataframe), reason about the output, and hand you a finished artifact in the editor. Because every number comes from a tool call, you can trace exactly how it was produced.
Yes. Every figure is retrieved from Alpha Sophia, not generated by the model, so the providers, NPIs, volumes, and market sizes that land in your code are real, current, and traceable to the underlying records — defensible enough to build commercial strategy and tooling on.
Access is authenticated per user and scoped to your organization, under the same enterprise-grade controls trusted by 100+ life science customers. Your org needs an active Alpha Sophia subscription with the API and Agent modules enabled. As with any AI workflow in a regulated environment, review outputs before they inform irreversible actions or regulated content.
Make informed connections with the most suitable healthcare providers for your product.