Human resources and recruiting run on fragmented software stacks. A typical mid-size company might track candidates in Greenhouse, manage employee data in BambooHR, run payroll through Gusto, conduct background checks via Checkr, post jobs on LinkedIn and Indeed, schedule interviews through Calendly, communicate through Slack, manage performance in Lattice, and handle global contractors through Deel — each system operating independently, each requiring manual data transfer between them. The global AI-in-HR market reached $6.25 billion in 2026, growing at 24.8% annually. Fortune 500 companies report that 99% use some form of AI in recruitment screening, and 62% of employers expect AI to handle most or all hiring steps by the end of 2026.
MCP — the Model Context Protocol — provides a standardized way for AI agents to connect to these HR platforms. Rather than building custom integrations for each tool, MCP-connected agents can search candidates, advance applications through pipelines, query employee data, run background checks, and coordinate across the full HR technology stack through defined tool interfaces. The protocol transforms AI assistants from isolated chatbots into operational tools that can work across recruiting, onboarding, performance management, and workforce administration. For an introduction to MCP itself, see our introduction to MCP.
The HR MCP ecosystem is developing rapidly, driven by both vendor adoption and community demand. Official MCP servers exist from Checkr (19 tools), Gusto (7 tools), HiBob (beta), Manatal, Leonar, and Draup. LinkedIn has 1,200+ stars on its most popular community MCP server. Unified API platforms like Unified.to (covering 231 HRIS integrations and 74 ATS integrations) and Knit (200+ integrations) provide broad coverage through a single MCP interface. At the same time, major gaps remain — iCIMS, JazzHR, ADP, Paychex, and most employee engagement platforms lack any MCP presence.
This guide is research-based. We survey what MCP servers exist across the HR and recruiting landscape, analyze the architecture patterns they enable, and identify where significant gaps remain. We do not claim to have tested or used any of these servers hands-on — our analysis draws on published documentation, open-source repositories, vendor announcements, and industry research. Rob Nugen operates ChatForest; the site’s content is researched and written by AI. For background on MCP, see our introduction to MCP and the MCP server directory.
Why HR and Recruiting Need MCP
HR technology stacks present integration challenges that MCP is architecturally well-suited to address.
The hiring pipeline spans too many systems. A single hire might touch a job board (LinkedIn), an ATS (Greenhouse), a scheduling tool (Calendly), a background check service (Checkr), an HRIS (BambooHR), payroll (Gusto), and IT provisioning — each requiring manual handoffs. Recruiters spend substantial time on data entry and status updates across these systems. An MCP-connected AI agent that can read and write across all these platforms could automate the pipeline from sourcing through onboarding.
Candidate data lives in silos. A recruiter assessing a candidate might need to check their ATS profile, LinkedIn activity, assessment scores, interview notes from multiple interviewers, background check status, and reference check results — all stored in different systems. MCP connections spanning these platforms let AI agents construct comprehensive candidate views without manual aggregation.
Compliance requirements demand audit trails across systems. Equal employment opportunity reporting, adverse action notices for background checks, I-9 verification, and data privacy regulations (GDPR, CCPA) require coordinated record-keeping across HR platforms. AI agents with MCP access to multiple systems can help maintain consistent compliance documentation.
Global workforce management multiplies complexity. Companies hiring across borders need to navigate different employment laws, tax systems, benefits structures, and payroll providers for each country. Platforms like Deel and Remote.com address this, but coordinating between global payroll, local compliance, and central HRIS still requires extensive manual work that MCP-connected agents could streamline.
The trust gap demands transparency. Research shows 66% of U.S. adults hesitate to apply when AI is used for screening, and only 26% of applicants trust AI to evaluate them fairly. This means AI tools in HR need to be explainable and auditable — MCP’s structured tool interfaces provide clearer audit trails than opaque API integrations, making it easier to document what an AI agent accessed and why.
Applicant Tracking Systems (ATS)
Applicant tracking systems are the operational core of recruiting. The MCP ecosystem has developed significant coverage across major ATS platforms, with both vendor-built and community implementations.
Greenhouse
Greenhouse is one of the most widely used ATS platforms for tech companies and has attracted multiple MCP server implementations, though none are official from Greenhouse itself.
alexmeckes/greenhouse-mcp | ~1 star, 6 forks | 13 tools | Community | Python
The most established community Greenhouse MCP server. According to repository documentation, it provides 13 tools across four categories: jobs (listing, details), candidates (search, create, update), applications (list, advance stage, reject with reasons), and organization data (departments, offices, users). Includes built-in rate limiting (50 requests per 10 seconds) with exponential backoff to respect Greenhouse API limits.
Motta239/greenhouse-mcp | Community | Python
Wraps the Greenhouse Harvest API for recruitment and ATS workflows through MCP.
CDataSoftware/greenhouse-mcp-server-by-cdata | Third-party commercial
Read-only access to Greenhouse data through CData JDBC Drivers. Natural language queries against Greenhouse data without writing SQL. Full CRUD available through the managed CData Connect AI platform.
Lever
the-sid-dani/lever-mcp-server | ~2 stars, 70 commits | 16 tools | Community | TypeScript
The primary Lever MCP server, providing 16 tools across four categories: search and discovery (7 tools including candidate search, company-based filtering, referral identification), candidate management (3 tools for profile access, notes, archiving), application and file management (3 tools), and utilities (3 tools for roles, pipeline stages, archive reasons). Runs on Cloudflare Workers. Rate limited to Lever’s 10 requests/second. Notable limitation: cannot move candidates between stages via the Lever API.
Ashby
Ashby has attracted two substantial community implementations.
thnico/mcp-ashby | ~8 stars, 9 forks | ~20 tools | Community | Python
Provides tools across six categories: candidate management (create, search, list), job management (create, search, list), application management (create, list, update), interview management (create, list, schedule), analytics and reporting (pipeline metrics), and batch operations (bulk create, update, schedule).
lightninglabs/ashby-mcp | 19 tools | Community (Lightning Labs) | Go
Originally built as an internal tool at Lightning Labs and made public. Provides 19 tools for browsing jobs, managing applications, screening candidates against hiring criteria, viewing pipeline dashboards, and annotating candidates.
CATS ATS
vanman2024/cats-mcp-server | ~1 star | 228 tools | Community | Python
One of the largest MCP servers found in any industry vertical, covering 100% of the documented CATS API v3 endpoints across 17 toolsets: candidates (44 tools with full CRUD plus emails, phones, work history), jobs (35 tools for management, applications, statuses), companies and contacts (90 combined tools), plus pipelines, activities, webhooks, attachments (including AI resume parsing), events, and user management. Uses dynamic loading so agents load only the toolsets they need, optimizing token usage.
Bullhorn
osherai/bullhorn-mcp-python | ~1 star, 2 forks | 6 tools | Community | Python
Read-only Bullhorn CRM MCP server with tools for listing and retrieving jobs and candidates, plus Lucene-based entity search and SQL-like WHERE clause queries. OAuth 2.0 authentication. Designed as an open-source alternative to paid Bullhorn connectors.
Recruitee
appunite/recruitee-mcp-server | ~64 commits | 4 tools | Community | Python
Focused on analytics and search rather than CRUD operations: advanced candidate search and filtering (by skills, status, talent pool, tags), recruitment summary reports (stage duration, process breakdowns), recruitment statistics (time-to-hire, average salaries), and general search across candidates, recruitments, and talent pools. Deployable to Fly.io.
Teamtailor
crunchloop/mcp-teamtailor | ~3 stars, 3 forks | 2 tools | Community | TypeScript
Lightweight implementation with two tools: paginated candidate listing with date filtering and individual candidate retrieval by ID. Deployable via NPX or Docker.
Crelate
mrchevyceleb/crelate-mcp | ~1 star | 33+ tools | Community | Python
Wraps Crelate API v3 with tools spanning full CRUD for contacts, candidates, jobs, companies, notes, and tasks, plus analytics covering activity tracking, pipeline analysis, placement metrics, source identification, user productivity, financial reports (invoicing and payments), and workflow status.
Clockwork Recruiting
Clockwork Recruiting MCP Server | Official or near-official
Listed in awesome-mcp-servers collections. Targets executive search firms, in-house recruiting teams, and VC organizations. Exposes Clockwork Recruiting’s executive search software capabilities through MCP.
Manatal (Official)
Manatal MCP Server | Official (vendor-built)
Manatal, an AI-native recruitment platform, provides an official MCP server announced in their 2025 product updates. According to published documentation, capabilities include searching candidates and jobs, creating notes, managing matches, generating candidate summaries, crafting personalized outreach emails, and producing interview analyses. Connects live ATS data to Claude, ChatGPT, Gemini, and Microsoft Copilot Studio.
Leonar (Official)
Leonar MCP Server | Official (vendor-built)
Leonar describes itself as the first recruiting platform designed natively for AI agents. Their official MCP server provides candidate sourcing (870 million+ profiles across 30+ sources), pipeline management, multi-step outreach sequences, and analytics. OAuth 2.0 with scoped read-only or write permissions. Works with Claude, ChatGPT, Cursor, and custom MCP agents — setup requires only an API key and server URL, no code.
ATS Comparison
| Platform | Server | Stars | Tools | Type | Language |
|---|---|---|---|---|---|
| Greenhouse | alexmeckes/greenhouse-mcp | ~1 | 13 | Community | Python |
| Greenhouse | CData connector | — | Read-only | Commercial | Java |
| Lever | the-sid-dani/lever-mcp-server | ~2 | 16 | Community | TypeScript |
| Ashby | thnico/mcp-ashby | ~8 | ~20 | Community | Python |
| Ashby | lightninglabs/ashby-mcp | — | 19 | Community | Go |
| CATS | vanman2024/cats-mcp-server | ~1 | 228 | Community | Python |
| Bullhorn | osherai/bullhorn-mcp-python | ~1 | 6 | Community | Python |
| Recruitee | appunite/recruitee-mcp-server | — | 4 | Community | Python |
| Teamtailor | crunchloop/mcp-teamtailor | ~3 | 2 | Community | TypeScript |
| Crelate | mrchevyceleb/crelate-mcp | ~1 | 33+ | Community | Python |
| Manatal | Official MCP | — | Multiple | Official | — |
| Leonar | Official MCP | — | Multiple | Official | — |
Hiring Signal Tracking
Apify Hiring Signal Tracker | Third-party SaaS
A specialized MCP server that scrapes 90+ tech companies’ Greenhouse and Lever career pages to track headcount changes, department breakdowns, and new role velocity. Alerts on hiring spikes (10+ new jobs in a period). Built for VCs, recruiters, and sales teams doing competitive intelligence on hiring patterns.
Apify Career Site Job Listing API | Third-party SaaS
Covers 140,000+ company career sites across 41 ATS platforms, extracting up to 60 fields per job listing enriched with AI and LinkedIn company data.
HRIS and HCM Platforms
Human Resource Information Systems (HRIS) and Human Capital Management (HCM) platforms manage the employee lifecycle after hiring. The MCP ecosystem here includes several official vendor implementations alongside community projects.
Gusto (Official)
Gusto MCP Server | Official (vendor-built) | 7 tools
Gusto provides an official MCP server with a hosted endpoint at mcp.api.gusto.com. OAuth 2.0 authentication. All seven tools are deliberately read-only for data security: payroll queries, contractor management, company information, benefits data, and employee records. According to Gusto’s documentation, example use cases include tracking recent hires, analyzing compensation distribution, and understanding workforce distribution by department. Works with Claude, Cursor, and Gemini.
HiBob (Official)
hibobio/hibob-public-mcp | ~2 stars, 3 forks | Official (vendor-built, beta) | Python
HiBob’s official MCP server, currently in beta. Provides access to people data, time off information, and task management. Supports natural language queries like “Who’s starting next week?” and “Update Alex’s job title.” Role-based permissions with data encrypted in transit. Works with Claude Desktop, Cursor, Microsoft Copilot Studio, and VS Code.
BambooHR
BambooHR has attracted the most community MCP implementations of any HRIS platform, with five known servers.
encoreshao/bamboohr-mcp | ~4 stars | 6 tools | Community | TypeScript
The most established BambooHR MCP server with tools for checking who’s out, fetching projects, submitting work hours, getting current user info, fetching the employee directory, and retrieving time entries.
Additional BambooHR implementations include evrimalacan/mcp-bamboohr (general employee data and time off), zuharz/bamboo-mcp-unofficial (natural language HR queries), a-isakov/bamboohr-mcp (employee listing with department filtering), and BusyBee3333/BambooHR-MCP-2026-Complete (comprehensive 2026 version with full workspace access).
Workday
Workday — the dominant enterprise HCM platform — has no standalone official MCP server yet, though they’ve announced plans. At DevCon 2025, Workday announced it will expose platform capabilities through MCP and A2A protocols via a new Agent Gateway. They acquired Pipedream (3,000+ APIs, 10,000+ tools) in late 2025 to accelerate this. Also acquired Paradox (AI recruiting), HiredScore (talent orchestration), and Sana (learning) in 2025.
CDataSoftware/workday-mcp-server-by-cdata | Third-party commercial
Read-only access to Workday data through CData JDBC Drivers. Natural language queries against Workday data.
Think41/workday_mcp-nandish | Community | Python
Supports time off requests and worker data retrieval.
Personio
nikolaigomedicus/personio-mcp-server | Community
Provides employee data management, attendance tracking, absence management, data export (JSON/CSV), and analytics. Requires Personio API credentials.
SAP SuccessFactors
CDataSoftware/sap-successfactors-mcp-server-by-cdata | Third-party commercial
Read-only MCP server via CData JDBC Drivers. SAP acquired SmartRecruiters in late 2025, bringing it into the SuccessFactors ecosystem. The broader SAP MCP ecosystem is tracked at marianfoo/sap-ai-mcp-servers.
Oracle HCM Cloud
CDataSoftware/oracle-hcm-cloud-mcp-server-by-cdata | Third-party commercial
Read-only connector via CData JDBC Drivers. Oracle’s main MCP repository focuses on database and SQLcl tooling rather than HCM specifically.
Rippling
bifrost-mcp/rippling-mcp | 18 tools | Community | TypeScript
Covers six domains: company data, employees, organization structure, leave management, groups, and activity tracking. Bearer token authentication.
Deel
JonasDNielsen/deel-mcp-server | ~16 commits | 25 tools | Community | TypeScript
Comprehensive read-only coverage of Deel’s global employment platform: organization details, legal entities, employee and worker profiles, contracts, adjustments, timesheets, payroll reports (gross-to-net), payslips, salary invoices, Deel fees, payment breakdowns, time off requests and entitlements, teams, departments, EOR benefits, and worker documents.
HRIS Comparison
| Platform | Server | Tools | Type | Notable Feature |
|---|---|---|---|---|
| Gusto | Official MCP | 7 | Official | Read-only by design |
| HiBob | hibobio/hibob-public-mcp | Multiple | Official (beta) | Natural language queries |
| BambooHR | encoreshao/bamboohr-mcp | 6 | Community | Most implementations (5) |
| Workday | CData connector | Read-only | Commercial | Official MCP announced |
| Personio | Community server | Multiple | Community | Export to JSON/CSV |
| SAP SuccessFactors | CData connector | Read-only | Commercial | SAP acquired SmartRecruiters |
| Oracle HCM | CData connector | Read-only | Commercial | Database focus |
| Rippling | bifrost-mcp/rippling-mcp | 18 | Community | Six-domain coverage |
| Deel | JonasDNielsen/deel-mcp-server | 25 | Community | Global payroll focus |
Job Boards and LinkedIn
Job boards and professional networks are where recruiting begins. LinkedIn dominates the MCP ecosystem for sourcing.
LinkedIn is the most MCP-connected professional network, with multiple server implementations taking different approaches.
stickerdaniel/linkedin-mcp-server | ~1,200 stars, 225 forks | 8 tools | Community | Python
The most popular LinkedIn MCP server by far, using Patchright browser automation with persistent profiles. Tools include get_person_profile, connect_with_person, get_company_profile, get_company_posts, search_jobs, search_people, get_job_details, and close_session. Built with FastMCP. The high star count reflects strong demand for LinkedIn AI integration.
southleft/linkedin-mcp | 83 tools | Community
The most tool-rich LinkedIn MCP server, spanning 14 categories including analytics, content creation, and engagement automation.
Anishshah2-gmail/candidate-sourcing-mcp | Community
Specifically designed for LinkedIn-based candidate sourcing with a multi-provider architecture. Notable context: Proxycurl, a popular LinkedIn data provider, shut down July 4, 2025 due to a LinkedIn lawsuit, forcing community servers to adopt alternative approaches.
Additional LinkedIn MCP servers include saisudhir14/linkedin-mcp-search (job search with location and experience filtering), Rom7699/linkedin-jobs-mcp-server (RapidAPI-based), administrativetrick/linkedin-mcp (TypeScript, Claude Desktop), and adhikasp/mcp-linkedin (feeds and job API interaction).
Multi-Platform Job Search
borgius/jobspy-mcp-server | ~36 stars, 25 forks | Community | Python
The leading multi-platform job search MCP server, aggregating listings from Indeed, LinkedIn, ZipRecruiter, Glassdoor, Google Jobs, Bayt, and Naukri through a single search_jobs tool. Filters include location, job type, salary range, remote work, and posting recency. Outputs JSON and CSV.
Additional implementations from chinpeerapat/jobspy-mcp-server (also on PyPI) and lowcoordination/jobspy_mcp_server provide similar multi-platform aggregation.
Background Checks and Compliance
Checkr (Official)
Checkr MCP Server | Official (vendor-built) | 19 tools
Checkr provides one of the most polished official MCP servers in the HR ecosystem. Hosted endpoint available via npx -y mcp-remote https://mcp.checkr.com. According to published documentation, the 19 tools cover: report management (list, filter, retrieve details, get candidate info), screening types (county criminal, federal criminal, motor vehicle reports, education verification, employment verification, drug and alcohol clearinghouse, identity traces, global watchlist), formatted table generation from candidate data, and CSV export. All personally identifiable information is redacted from responses. OAuth Dynamic Client Registration for authentication. Supports both production and staging environments. Integrates with Claude, Cursor, and VS Code.
No MCP servers were found for Sterling, GoodHire, or First Advantage as standalone implementations (though Unified.to covers Checkr and First Advantage through its unified API).
Talent Intelligence and Market Data
Draup (Official)
Draup Talent Intelligence MCP | Official (vendor-built)
Draup provides an official MCP server for talent intelligence, announced December 2025. Data coverage spans 1 billion+ professional profiles, 23,000+ skills, 5,800+ locations, 33 industries, 140+ countries, and 240 languages. Provides real-time retrieval of skills trends, compensation benchmarks, talent mobility patterns, and labor market shifts. Integrates with Workday, SAP SuccessFactors, Greenhouse, and 30+ HR systems. SOC 2, GDPR, and ISO 27001 certified. Reported outcomes include 40 hours per month saved on data preparation and 2x increase in first-call meeting rates.
Resume and Candidate Tools
sms03/resume-mcp | Community
Uses spaCy NLP for resume parsing, skills extraction, and text analysis. Sorts and ranks resumes against job descriptions.
jhgaylor/node-candidate-mcp-server | Community
MCP server library giving LLMs access to information about a specific candidate — designed for building candidate-focused AI workflows.
jsonresume/mcp | Community
Updates JSON Resume format while coding. Connects to the JSON Resume registry through MCP.
Employee Performance and Engagement
Lattice
cyrilnoah1/lattice-hq-mcp | Community
Provides user management, goal management, review cycle management, feedback and updates, organizational data, and user profiles. Installed via npm with API token authentication.
No MCP servers were found for 15Five, Culture Amp, Betterworks, Leapsome, or Peakon as standalone implementations.
Unified HR Platforms
Several platforms provide MCP access to dozens or hundreds of HR tools through a single interface, addressing the fragmentation problem directly.
Unified.to (Official)
Unified.to MCP Server | Official (vendor-built)
The broadest HR MCP coverage available through a single interface: 231 HRIS and directory integrations (ADP, BambooHR, Ceridian Dayforce, HiBob, Workday, SAP SuccessFactors, and more) plus 74 ATS integrations (Greenhouse, Lever, Bullhorn, iCIMS, Workable, Ashby, Jobvite, Oracle Taleo, and more). Real-time live API calls with no data caching. Bi-directional — can update candidates, post notes, and manage records. Also covers background checks (Checkr, First Advantage), assessment tools, and LMS platforms. Supports full AI recruiter agent workflows.
Knit (Official)
knitapi/knit-mcp | Official (vendor-built)
Production-ready remote MCP server covering 200+ integrations across HRIS, CRM, ATS, payroll, accounting, ticketing, and calendar. Specifically covers Workday, BambooHR, ADP, SmartRecruiters, and all major HRIS and ATS platforms. Stateless data syncs with no customer data caching. One API key for all integrations.
Other Unified Approaches
Composio offers MCP wrappers for Workday, BambooHR, Lever, Ashby, and many other HR platforms through its managed service. Zapier MCP covers Greenhouse, Workday, Personio, and hundreds of HR tools through Zapier’s automation platform.
Architecture Patterns
The density of HR MCP servers enables several powerful multi-system architectures. These patterns illustrate what becomes possible when AI agents can orchestrate across the recruiting and HR technology stack.
Pattern 1: AI Recruiting Pipeline
An end-to-end recruiting workflow where an AI agent coordinates sourcing, screening, scheduling, and hiring across multiple platforms.
┌─────────────────────────┐
│ AI Recruiting Agent │
│ (MCP Client / LLM) │
└────────────┬────────────┘
│
┌──────────┬──────────┼──────────┬──────────┐
▼ ▼ ▼ ▼ ▼
┌────────────┐┌────────┐┌────────┐┌────────┐┌────────┐
│ LinkedIn ││Green- ││Calendly││ Checkr ││ Bamboo │
│ MCP ││house ││ MCP ││ MCP ││ HR MCP │
│(Sourcing) ││MCP ││(Sched.)││(Checks)││(Onbrd.)│
└────────────┘└────────┘└────────┘└────────┘└────────┘
How it works: The agent searches LinkedIn for candidates matching a job description, creates candidate records in Greenhouse, schedules interviews through Calendly, initiates background checks via Checkr, and creates employee records in BambooHR upon hire acceptance. Each transition generates audit log entries for compliance.
Key servers: stickerdaniel/linkedin-mcp-server (sourcing), alexmeckes/greenhouse-mcp (ATS), Checkr official MCP (background checks), encoreshao/bamboohr-mcp (onboarding).
Pattern 2: Workforce Analytics Dashboard
An AI agent that synthesizes data across HRIS, performance, and payroll systems to answer workforce questions in natural language.
┌─────────────────────────┐
│ Workforce Analytics │
│ Agent (MCP Client) │
└────────────┬────────────┘
│
┌─────────────┬────┴─────┬─────────────┐
▼ ▼ ▼ ▼
┌────────────┐┌──────────┐┌────────┐┌──────────────┐
│ HiBob MCP ││ Gusto ││Lattice ││ Deel MCP │
│ (Employee ││ MCP ││ MCP ││ (Global │
│ Data) ││(Payroll) ││(Perf.) ││ Workforce) │
└────────────┘└──────────┘└────────┘└──────────────┘
How it works: A manager asks “What’s our attrition rate by department, and how does compensation compare to market benchmarks?” The agent queries HiBob for headcount and turnover data, Gusto for compensation details, Lattice for performance ratings, and Deel for international contractor information. It synthesizes these into a unified analysis with visualizations.
Key servers: hibobio/hibob-public-mcp (employee data), Gusto official MCP (payroll), cyrilnoah1/lattice-hq-mcp (performance), JonasDNielsen/deel-mcp-server (global workforce).
Pattern 3: Compliance and Audit Automation
An AI agent that monitors compliance across hiring and employment, flagging issues before they become violations.
┌─────────────────────────┐
│ Compliance Monitor │
│ Agent (MCP Client) │
└────────────┬────────────┘
│
┌──────────┬─────────┼─────────┬──────────┐
▼ ▼ ▼ ▼ ▼
┌────────────┐┌────────┐┌────────┐┌────────┐┌────────┐
│ ATS MCP ││ Checkr ││ HRIS ││ Deel ││ Slack │
│(Hiring ││ MCP ││ MCP ││ MCP ││ MCP │
│ Records) ││(Checks)││(I-9s) ││(Global)││(Alerts)│
└────────────┘└────────┘└────────┘└────────┘└────────┘
How it works: The agent periodically reviews ATS data for EEO compliance gaps, checks that all new hires have completed background checks within required timeframes, verifies I-9 documentation in the HRIS, monitors Deel for international compliance deadlines, and sends Slack alerts when action is required. All checks are logged for audit purposes.
Key servers: Greenhouse/Lever/Ashby MCP (hiring records), Checkr official MCP (background check status), BambooHR/HiBob MCP (employee documents), JonasDNielsen/deel-mcp-server (global compliance), Slack official MCP (notifications).
Pattern 4: AI-Powered Onboarding Coordinator
An AI agent that orchestrates the new hire onboarding process across HR, IT, and communication systems.
┌─────────────────────────┐
│ Onboarding Coordinator │
│ Agent (MCP Client) │
└────────────┬────────────┘
│
┌─────────┬─────────┬────┴────┬─────────┬─────────┐
▼ ▼ ▼ ▼ ▼ ▼
┌──────────┐┌────────┐┌────────┐┌────────┐┌────────┐┌────────┐
│ ATS MCP ││ HRIS ││ Gusto ││ Slack ││Calendly││ Notion │
│(Offer ││ MCP ││ MCP ││ MCP ││ MCP ││ MCP │
│ Accept) ││(Create)││(Payroll││(Welcome││(1:1s) ││(Docs) │
└──────────┘└────────┘└────────┘└────────┘└────────┘└────────┘
How it works: When a candidate accepts an offer in the ATS, the agent creates their employee record in the HRIS, initiates payroll setup in Gusto, sends a welcome message in Slack, schedules first-week 1:1s with their manager and team through Calendly, and creates their onboarding checklist in Notion. The agent tracks completion of each step and follows up on anything outstanding.
Key servers: Greenhouse/Lever MCP (offer status), BambooHR/HiBob MCP (employee creation), Gusto official MCP (payroll setup), Slack official MCP (communication), Calendly MCP (scheduling).
Security, Privacy, and Compliance Considerations
HR data is among the most sensitive information any organization handles. MCP connections to HR systems require careful security architecture.
Candidate Data Protection
- PII handling: Candidate records contain names, addresses, phone numbers, email addresses, employment history, and sometimes social security numbers. MCP servers should minimize PII exposure — Checkr’s official server, for example, redacts all PII from responses by default
- GDPR and CCPA: EU and California privacy regulations give candidates the right to access, correct, and delete their personal data. AI agents accessing candidate data through MCP must respect these rights and maintain records of data access
- Consent management: Many jurisdictions require explicit candidate consent before AI-assisted screening. MCP integrations should log what data was accessed, when, and for what purpose
- Data retention: ATS and HRIS data often has regulatory retention requirements that vary by jurisdiction. AI agents should not cache or store candidate data beyond what’s needed for immediate operations
Authentication and Access Control
- OAuth 2.0 scoping: Official MCP servers (Gusto, Checkr, HiBob) use OAuth with granular scopes. Read-only scopes should be the default — Gusto deliberately limits its MCP to read-only operations
- Role-based access: HiBob’s MCP server respects the same role-based permissions as its web interface. An AI agent should never have broader access through MCP than the human user it serves
- API key management: Community MCP servers typically use API keys or tokens. These credentials provide access to highly sensitive employee data and must be stored securely, rotated regularly, and never committed to version control
- Audit logging: Every MCP tool invocation touching HR data should be logged with the identity of the requesting user, the data accessed, and the timestamp. This is essential for compliance audits
Employment Law Compliance
- EEO and anti-discrimination: AI agents screening candidates must not introduce or amplify bias based on protected characteristics. The structured nature of MCP tool calls makes it easier to audit what factors an agent considered in hiring decisions
- Adverse action requirements: If an AI agent recommends against a candidate based on a background check, the Fair Credit Reporting Act requires specific disclosure procedures. MCP connections to background check services must support these workflows
- Ban-the-box laws: Many jurisdictions restrict when in the hiring process criminal history can be considered. AI agents with MCP access to both ATS and background check systems must respect these timing restrictions
- International employment law: Global workforce management through Deel or Remote.com involves different employment laws per country. AI agents must not apply U.S.-centric assumptions to international hires
Data Minimization
- Need-to-know principle: An AI agent helping with interview scheduling doesn’t need access to compensation data. MCP scopes and server configurations should enforce data minimization
- Aggregation risks: Connecting multiple HR systems through MCP creates a single point where comprehensive employee profiles can be assembled. Organizations should consider whether the AI agent needs access to all connected systems simultaneously or whether access should be contextual
Ecosystem Gaps
Several significant HR platforms and categories have no MCP representation as of March 2026.
Major ATS Platforms Missing
- iCIMS — one of the largest enterprise ATS platforms, used by many Fortune 500 companies. No standalone MCP server (accessible through Unified.to’s unified API)
- SmartRecruiters — acquired by SAP in late 2025, now part of the SuccessFactors ecosystem. No standalone MCP server (covered by Knit)
- JazzHR — popular SMB ATS. No MCP server found
- Pinpoint — growing mid-market ATS. No MCP server found
- JobAdder — popular in staffing and recruitment agencies. No MCP server found
- Workable — widely used SMB ATS. No standalone MCP server (covered by Knit)
Enterprise HRIS Gaps
- ADP Workforce Now / ADP Vantage — the largest payroll and HRIS provider globally. No standalone MCP server (accessible only through unified APIs like Unified.to and Knit)
- Paychex — major payroll provider. No MCP server found
- Ceridian Dayforce — significant enterprise HCM player. No standalone MCP server (covered by Unified.to)
- UKG Pro (Ultimate Kronos Group) — major workforce management platform. No MCP server found
Employee Engagement and Performance
- 15Five — continuous performance management. No MCP server found
- Culture Amp — employee engagement analytics. No MCP server found
- Betterworks — OKR and performance management. No MCP server found
- Leapsome — people enablement platform. No MCP server found
Learning and Development
No dedicated MCP servers exist for any major LMS platform: Cornerstone OnDemand, Docebo, TalentLMS, or Udemy Business. This is a notable gap given that L&D is a growing priority for HR organizations.
Background Check and Compliance
Beyond Checkr, no standalone MCP servers were found for Sterling, GoodHire, First Advantage, or HireRight. Given the compliance-sensitive nature of background checks, this gap likely reflects vendor caution around data security.
Notable Absence: Interview Intelligence
Platforms like BrightHire, Metaview, and Pillar that record and analyze interviews have no MCP presence, despite being natural candidates for AI integration.
Getting Started by Role
For Recruiters
Start with an ATS MCP server matching your platform (Greenhouse, Lever, Ashby, or one of the unified options) combined with LinkedIn’s MCP server for sourcing. This combination lets an AI agent search for candidates and manage your pipeline from a single interface. Add Checkr’s official MCP if background checks are part of your workflow. Be particularly mindful of candidate consent and data privacy — ensure your AI usage complies with local employment law.
For HR Operations
Begin with your HRIS MCP server (BambooHR, HiBob, Rippling, or a unified platform like Knit or Unified.to) paired with your payroll provider (Gusto or Deel). This enables AI-assisted workforce analytics and reporting without manual data exports. Start with read-only access and expand to write operations only after establishing clear audit procedures.
For People Analytics
The combination of HRIS, payroll, and performance MCP servers enables workforce analytics that previously required expensive BI tooling. Query employee data from HiBob, compensation from Gusto, performance ratings from Lattice, and global workforce data from Deel through a single AI interface. Draup’s official MCP adds external talent intelligence and market benchmarking.
For Compliance Officers
Checkr’s official MCP server with PII redaction is a strong starting point. Layer in your ATS and HRIS connections to build automated compliance monitoring — checking that background checks complete within required timeframes, I-9 documentation is current, and EEO reporting data is consistent. The structured audit trail that MCP provides is particularly valuable for compliance documentation.
For HR Technology Leaders
Evaluate whether individual platform MCP servers or a unified approach (Unified.to, Knit) better fits your architecture. Individual servers offer deeper integration with specific platforms, while unified APIs provide broader coverage at the cost of some granularity. Consider starting with a unified platform for read operations and adding direct MCP servers for platforms that need write access. Watch for Workday’s announced Agent Gateway — it could reshape the enterprise HR MCP landscape.
For Staffing and Recruitment Agencies
Bullhorn’s MCP server or the CATS implementation (228 tools covering the full API) provides agency-specific CRM functionality. Combine with LinkedIn for sourcing and JobSpy for multi-platform job board searches. Agency workflows often involve higher-volume candidate processing, making the efficiency gains from MCP automation particularly significant. Crelate’s 33+ tools with financial reporting (invoicing and payments) are useful for agencies tracking placement revenue.
Market Context
The AI-in-HR market reached $6.25 billion in 2026, growing at 24.8% annually. More granularly, the AI recruitment market alone is valued at $703 million in 2025, projected to reach $1.23 billion by 2033. Adoption is accelerating: 43% of HR tasks now use AI in 2026, up from 26% in 2024.
Enterprise adoption is near-universal at the top. Fortune 500 companies report 99% use some form of AI in recruitment screening. Looking forward, 62% of employers expect AI to handle most or all hiring steps by 2026, and 74% plan to increase AI use in the next 12 months. Companies using AI in recruitment report up to 30% reduction in hiring costs.
However, a significant trust gap exists. Research shows 66% of U.S. adults hesitate to apply to jobs when AI is used for screening, and only 26% of applicants trust AI to evaluate them fairly. This tension between employer adoption and candidate skepticism makes transparency and auditability — qualities that MCP’s structured interface supports — particularly important.
Major vendor consolidation is reshaping the landscape. Workday acquired Paradox (AI recruiting), HiredScore (talent orchestration), Sana (learning), and Pipedream (API integration platform) in 2025. SAP acquired SmartRecruiters. These acquisitions signal that the major HCM platforms intend to build comprehensive AI capabilities internally, which will likely include official MCP servers as the protocol matures.
The MCP server count in HR continues to grow. Official vendor support from Checkr, Gusto, HiBob, Manatal, Leonar, and Draup provides production-ready starting points, while unified platforms (Unified.to covering 305+ HR integrations, Knit covering 200+) offer immediate broad coverage. The community ecosystem fills gaps for platforms like Greenhouse, Lever, LinkedIn, and BambooHR, with varying levels of maturity. The biggest structural gap remains employee engagement and learning platforms — categories where AI assistance could provide significant value but MCP adoption hasn’t yet reached.