Resources · Claude connector

Claude Enrichr connectorWhat Claude can do with Enrichr.

The Claude Enrichr connector exposes 2 tools, both read. 2 read, 0 write, 0 undocumented: Claude suggests the right Enrichr libraries, runs a gene set enrichment on your list and explains the significant terms. Here is what it does, where it stops, and how it installs.

Verified Trustpilot reviews · AI, automation & growth agency

Overview

Enrichment results you can talk through

A biologist with a list of hits usually pastes it into the Enrichr website, ticks libraries one by one and exports tables. With this desktop extension, you give the list to Claude. It can first find the libraries that fit your question, then query several of them in parallel and walk you through the terms that come out as significant.

Choosing libraries for a new topic. Working on DNA repair for the first time, a postdoc asks which Enrichr collections make sense. Claude calls suggest_libraries, which searches Enrichr's live catalogue for the best fit.

Running the analysis itself. With the gene list and the chosen libraries, enrichr_analysis queries them in parallel and returns only terms that pass the corrected p-value threshold, with odds ratios and overlapping genes.

Correcting for your real background. When the genes come from a targeted panel or one tissue, you can pass that universe as background, which the author says avoids overstating significance against the whole genome.

What it will not do: both tools read and analyse; neither writes into an account, and nothing runs on its own. Claude only calls them while it answers you. For a pipeline that reruns enrichment on every new dataset, an automation platform is the better fit, and the Integrations hub lists the ones we cover.

Vocabulary

Five words before you install

The Claude terms this page relies on, in a minute.

Connector
The link you set up once between Claude and an outside service, so Claude can use that service while it answers you.
Tool
One named job a connector opens to Claude. Claude decides by itself which ones to call; the directory sheet lists them by name.
Authorization
Usually the sign-in screen of the service, where you hand Claude its access. Enrichr's author documents no such screen here.
Approval
The go-ahead Claude waits for before it changes something in an account, shown in the conversation when it matters.
MCP
The shared standard connectors are built on: it is what lets an assistant like Claude talk to an outside service such as Enrichr.
Connect

Set up the Claude Enrichr connector in three steps

  1. 01

    Open the connectors in Claude Desktop

    The directory lists Enrichr as a desktop extension that works with Claude Desktop only. Open its settings, go to Customize then Connectors, and look for Enrichr there.

  2. 02

    Add it from its row

    No Claude source describes adding this extension from the list: the only documented route is the author's, in the tip below. No source describes an Enrichr sign-in either: the author documents no account step.

  3. 03

    Check who publishes it

    Before relying on it, note that the directory names an individual developer, Tianqi Tang, as publisher. Anthropic does not vouch for the tools a publisher exposes, so judge it as you would any research script.

Tools

The 2 tools, and what each one returns

Enrichr MCP Server gives Claude 2 tools: 2 that read your account, 0 that change something in it.

Two tools, both described in the author's README. Names stay exactly as Claude shows them.

  • 2 read

What Claude reads (2)

2 tools

Two tools that look up Enrichr libraries and run enrichment, without touching any account.

enrichr_analysis

Runs your gene symbols against several Enrichr libraries at once and brings back the significant terms with their p-values, odds ratio, combined score and the genes that overlap. An optional background list corrects the statistics.

When it helps
after an RNA-seq screen, you want the pathways your 40 hits point to, without clicking through each library.
Watch out
a list needs at least 2 genes, and a custom background at least 20.

Sourcegithub.com · October 1, 2026 ↗

suggest_libraries

Finds the Enrichr libraries that best fit a research question, drawn from the live catalogue so a retired library is never suggested. The author advises calling it before the analysis.

When it helps
a student starting on breast cancer drug resistance needs to know which of the many collections to query first.
Watch out
it returns 10 suggestions by default and caps at 50, so a vague topic can bury the useful one.

Sourcegithub.com · October 1, 2026 ↗

Approvals

What Claude asks before acting

By default, Claude stops and asks for your go-ahead before any action it takes on an account for you. The request appears in the conversation, at the moment it matters.

On Team and Enterprise workspaces, owners decide whether a member may let some actions through without a new prompt each time. They can also cap what a connector may do for the whole organization, for example keep reading and close writing, and nobody overrides that from their own account. Claude works with the access of the person who installed it and nothing more. Here, both tools only query Enrichr and return results.

Plans

Which plans include it

None of the 819 sheets in the official directory displays availability by plan. Connector by connector, that answer is not published anywhere: it is a real gap in the catalogue.

What is published is the general rule: remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile, while desktop extensions like this one install on Claude Desktop. On Team and Enterprise, an Owner or Primary Owner turns a connector on for the organization first. Check the sheet in the official directory for the current status. Only the author's README documents this extension.

Limits

Where this connector stops

A connector is not an automation. Claude calls these tools while it answers you; nothing reruns an analysis when a new dataset lands.

The tool list is an observed floor, not a promise. The directory badge is not a security audit either, and Anthropic states on every sheet that it does not choose the tools a publisher exposes or guarantee their behavior. Here the publisher is one individual developer, which is worth knowing before you build a figure on it. No source says either whether Claude asks for confirmation before enrichr_analysis or suggest_libraries.

Need help

Need help connecting Enrichr MCP Server to Claude?

A person reads every message.

FAQ

Questions about the Claude Enrichr connector

01What can Claude do with the Enrichr connector?
Claude can run gene set enrichment analysis with Enrichr from a conversation. One tool suggests the libraries that fit a research question, searching the live catalogue and never suggesting a retired library. The other sends your gene list to several libraries in parallel and returns the significant terms with p-values, odds ratios and overlapping genes. Claude then reads those results with you, and explains which pathways stand out. The author also documents a guided workflow that walks through library selection, analysis and interpretation.
02Can Claude write or delete anything through Enrichr?
Neither tool writes into an account: both query Enrichr and return results, and by default the sheet only exposes reading. The one thing the author documents on the output side is an option that saves the complete results as a TSV file on your own computer. That remains an observed floor rather than a promise, since a public sheet never shows everything an administrator might open. No source describes a write tool on this connector.
03Does Claude ask before running an enrichment?
No official source describes a confirmation specific to these two tools, so Claude's default rule applies: it asks for your approval in the conversation before any action it takes on an account for you. On Team and Enterprise workspaces, owners decide whether members may let some actions through without a fresh prompt. They can also restrict what a connector may do for the whole organization, and members cannot override that from their own settings.
04Which Claude plans is the Enrichr connector available on?
No official source publishes plan availability connector by connector, and none of the 819 directory sheets displays it. The general rule is that remote connectors are open to all users on Claude, Cowork, Claude Desktop and mobile, and that desktop extensions install on Claude Desktop, with an Owner or Primary Owner enabling connectors first on Team and Enterprise. The author's README names no Claude plan. The directory sheet shows the current status.
05Does Claude see my data when it uses Enrichr?
Claude sees what you give it in the conversation, typically your gene list, and what Enrichr sends back. The extension has no account documented by its author, so there is no personal library for Claude to browse. Your genes go to the Enrichr service for analysis, and background correction goes through a separate Enrichr service. On the Claude side, an owner of a Team or Enterprise workspace can still restrict what any connector may do.
06Why do my results say background correction failed?
Background correction runs on a separate Enrichr service that the author describes as intermittently unavailable. The extension retries; if it still fails, that library falls back to uncorrected whole-genome p-values and the result is flagged with a clear warning, never presented as corrected. In that case, rerun later or read those p-values with care, since the whole-genome default can overstate significance for a restricted gene universe. A custom background must hold at least 20 genes.
07Claude or an automation tool for Enrichr?
They suit different jobs, so pick by task rather than by ranking. Claude fits exploratory work in a conversation: choosing libraries for a new question, running one analysis and discussing what the terms mean. It only acts while you are talking to it. An automation platform or a scripted pipeline fits repeated runs that happen without you, such as analysing every new screen the same way. Nothing in this connector runs on its own between two conversations.