Claude Enterprise
Claude for HR: what it does and where a person must decide
Anthropic publishes an HR plugin for Claude and a usage policy that lists employment decisions as high risk. This page reports both and says what we could not read.
Claude for HR, on Anthropic's own pages, means Claude doing people operations work through an Anthropic published Human Resources plugin for recruiting, onboarding, performance reviews, compensation analysis and policy guidance. Anthropic's usage policy lists employment decisions, including hiring, promoting, setting pay and evaluating a worker's performance, as a high risk area where a qualified person must review what Claude recommends. Source: Anthropic plugin page, GitHub README and usage policy, read 18 November 2026.
plain answer
What the Human Resources plugin does
Anthropic's plugin page describes the Human Resources plugin as work on "recruiting, onboarding, performance reviews, compensation analysis, and policy guidance." The page marks it "Anthropic verified." The README in Anthropic's knowledge-work-plugins repository gives the same list in its own words: "Helps with recruiting, onboarding, performance management, policy guidance, and compensation analysis."
The plugin page gives four example requests: "Draft a performance review summary based on this quarter's goals and feedback", "Benchmark our senior engineer compensation against market data", "Create a 30-60-90 day onboarding plan for a new product manager" and "Summarize our PTO policy and how it applies to part-time employees".
plain answer
The six commands and six skills in the README
The README lists six commands you start with a slash. /draft-offer drafts an offer letter with compensation details, start date and terms. /onboarding generates an onboarding checklist and first-week plan for a new hire. /performance-review structures a performance review: self-assessment prompts, a manager template and calibration prep. /policy-lookup finds and explains company policies such as PTO, benefits, expense, travel and remote work. /comp-analysis analyzes compensation data: benchmarking, band placement and equity refresh modeling. /people-report generates headcount, attrition, diversity or org health reports.
It also lists six skills, described as "Domain knowledge Claude uses automatically when relevant": recruiting-pipeline, employee-handbook, compensation-benchmarking, org-planning, people-analytics and interview-prep. The README says interview-prep creates structured interview plans with competency-based questions, scorecards and debrief templates, and that recruiting-pipeline tracks the source, screen, interview and offer stages.
plain answer
Where the plugin runs and which plans include Cowork
The README calls it "A people operations plugin primarily designed for Cowork, Anthropic's agentic desktop application" that "also works in Claude Code." It says every command and skill "works without any integrations", and describes the plugin as working "standalone with your input, supercharged when you connect your HRIS, ATS, and other tools."
Anthropic's Cowork page says plugins "bundle your tools, knowledge, and workflows into a single install so Claude shows up as a specialist from day one." On the plan cards it says "Claude Cowork is included in your Pro plan", "Claude Cowork is included in your Max plan", "Even more Claude Cowork included in your Ultra plan" and "Claude Cowork is included in your Enterprise plan", and that "Claude Cowork and the Slack connector are included in your Team plan standard and premium seats." A banner on the page says "Claude Cowork is now just Claude" and that it is "Rolling out to Pro, Max, and Ultra, with more plans to follow." We did not open the page the banner links to, so we do not say what the change means for the HR plugin.
For Claude Code, the README's install steps are a marketplace add command and a plugin install command; we did not run them. Anthropic's Claude Code documentation says of any plugin: "A plugin can run hooks and MCP servers, so read the pane before you install."
plain answer
What the plugin can connect to
The README groups the optional connections into six kinds of tool: HR information systems, applicant tracking systems, compensation data, chat, calendar and email. Anthropic's connectors file for the plugin adds a seventh, a knowledge base, and shows which kinds already have a server included. Calendar, chat, email and knowledge base list included servers. HR information systems, applicant tracking systems and compensation data list none, only other options. The file says "any MCP server in that category works."
Without a connection, the README's table says to give the plugin the material yourself: provide offer details manually, describe your onboarding process, upload a CSV or describe your bands for compensation analysis, and paste handbook content for policy lookup. With connections, it says the plugin can fill offers from an HR system, read review history, and use live data for people reports.
Our reading, not Anthropic's: employee records and candidate files are among the most sensitive data a company holds, so which plan holds them, who can reach them and what the retention settings are belong in a written decision before the first connection, not after.
plain answer
What Anthropic says about stored conversations
Anthropic's privacy article on how long it stores an organization's data was last dated July 1, 2026. For its commercial products that let you save and continue conversations, it says: "we retain your chats and coding sessions in the product to provide you with a consistent product experience." It says you control your conversations and can delete your chats at any time, and that a deleted conversation is removed from your chat history immediately and deleted from Anthropic's back-end storage systems within 30 days.
The article also says that if a chat is flagged by Anthropic's automated trust and safety systems as violating its Usage Policy, inputs and outputs are kept for up to 2 years, and that Anthropic may retain chats or sessions as required by law. It does not mention Cowork or plugins, so we do not say how it applies to data the HR plugin handles. It lists a separate article on custom data retention controls for Enterprise plans, which we did not open. For the administrator side, see our page on the Claude admin console.
plain answer
Anthropic lists employment as a high risk area
Anthropic's usage policy lists employment under its high risk areas: "screening, ranking, advancing, or rejecting candidates; hiring; promoting or demoting; setting or changing shifts, hours, or pay; evaluating a worker's performance or behavior, including through disciplining, suspending, or terminating a worker."
For those uses the policy says "A qualified person must meaningfully review a High-risk AI Recommendation, with the authority to change or modify it," before it is used to implement a decision. It defines a qualified person as "someone with the training or experience to evaluate the output in that field, and who holds a license where applicable law requires one for the underlying activity." It adds that "The qualified person remains responsible for the accuracy and appropriateness of what is delivered or decided."
On disclosure, the policy says "An individual who receives advice or who is the subject of a decision based on a High-risk AI Recommendation must be clearly told that AI was used to produce it," and "Users do not need to identify Anthropic, Claude, or the model used." It also says these requirements are meant to protect individuals but "do not replace local legal requirements and restrictions."
plain answer
Which plugin jobs sit near that list
Drafting a job description, an onboarding plan or a policy summary is work a person reviews before it goes out. Work that feeds a decision about a candidate or a worker's pay or performance is on Anthropic's high risk list. The policy also says its requirements do not apply to "Internal drafting, research, summarization, or analysis that is not the ultimate High-risk AI Recommendation delivered" or to "Business operations that do not advise or decide about a specific individual."
Our reading, not Anthropic's: of the six commands, /performance-review and /comp-analysis sit closest to the Employment line, because a review or a pay band ends up attached to a named worker, and /draft-offer sets pay for a named candidate. Whether a particular use falls inside or outside the exclusions is a question for your counsel, not this page.
Our own course for HR and recruiting teams, linked below, covers the drafting side with the decisions kept human. It is not specific to Claude.
limits
What we did not read
We did not install or test the plugin, and we read the README's description of the commands and skills, not the plugin's own skill files. We did not open the page the Cowork banner links to, Anthropic's article on custom data retention controls for Enterprise plans, or any Anthropic page on how plugin data is handled. Anthropic's tutorial page "Claude for Human Resources" would not open for us (see Cited sources), so nothing here relies on it. Kriv AI is not a law firm and this page is not legal advice; your HR lead and counsel decide whether a use meets your obligations.
Sources
Cited sources
- S1: Human Resources plugin (Claude Marketplace) (no publication or update date shown; footer © 2026 Anthropic PBC), read 2026-11-18
- S2: human-resources plugin README (anthropics/knowledge-work-plugins) (none shown (raw README.md read with curl, HTTP 200)), read 2026-11-18
- S3: Usage Policy (Anthropic) (effective date shown: Nov 12, 2026), read 2026-11-18
- S4: Claude Cowork (claude.com product page) (no publication date shown; footer © 2026 Anthropic PBC), read 2026-11-18
- S5: Connectors file for the human-resources plugin (CONNECTORS.md) (none shown (raw file read with curl, HTTP 200)), read 2026-11-18
- S6: Discover and install prebuilt plugins (Claude Code docs) (no date shown), read 2026-11-18
- S7: How long do you store my organization's data? (Anthropic Privacy Center) (article dated July 1, 2026), read 2026-11-18
Straight answers
Frequently asked questions about Claude for HR: what it does and where a person must decide
Is there a Claude plugin for HR?
Yes. Anthropic's plugin page lists a Human Resources plugin marked "Anthropic verified" for recruiting, onboarding, performance reviews, compensation analysis and policy guidance. Source: S1, read 18 November 2026.
Can Claude be used for hiring, pay or performance decisions?
Anthropic's usage policy lists employment decisions as a high risk area, including "screening, ranking, advancing, or rejecting candidates", setting or changing pay, and evaluating a worker's performance. It says a qualified person must meaningfully review the recommendation. Whether your use is allowed is for your counsel. Source: S3, read 18 November 2026.
Which Claude plans include Cowork, where the HR plugin runs?
Anthropic's Cowork page says it is included in the Pro, Max, Ultra and Enterprise plans, and that Cowork and the Slack connector are included in Team plan standard and premium seats. A banner on the page says Cowork is rolling out to Pro, Max, and Ultra, with more plans to follow. Source: S4, read 18 November 2026.
Does the HR plugin need my HR systems connected?
No. The README says every command and skill works without any integrations, and that the plugin works "standalone with your input, supercharged when you connect your HRIS, ATS, and other tools." Source: S2, read 18 November 2026.
Do I need to tell candidates or employees that Claude was used?
Anthropic's usage policy says an "individual who receives advice or who is the subject of a decision based on a High-risk AI Recommendation must be clearly told that AI was used to produce it," and that users "do not need to identify Anthropic, Claude, or the model used." Whether and how that applies to you is for your counsel. Source: S3, read 18 November 2026.
Does Anthropic keep my conversations?
For its commercial products that save conversations, Anthropic's privacy article says it retains chats in the product, that you can delete them, and that deleted conversations are removed from its back-end storage systems within 30 days, with exceptions for flagged chats and legal requirements. The article does not mention Cowork or plugins. Source: S7, read 18 November 2026.
Talk to the team that would do the work
Bring your requirements to a working session with the person who'll actually deliver.
Book a Discovery Call