NAPPLIED MCP SERVER · BETA
Run hiring from the AI assistant you already use.
Recruiters already ask Claude and ChatGPT to write Boolean strings and job descriptions, but a chatbot cannot see your applicants, your interviews or your pipeline. The Napplied MCP server gives it that access, safely, so the assistant can find, rank, interview and move candidates for you.
https://napplied.com/employer/mcp- Add the server URL in your assistant as a custom connector.
- It opens Napplied. Sign in and choose Full access or Read-only.
- Ask in plain English. Anything that emails a candidate shows you a preview first.
WHAT YOU CAN ASK
From questions to a full sourcing agent.
Ask about your pipeline
Recruiters and hiring managers
- Who are the top applicants for the Senior Backend Engineer role, and why?
- Summarise the Morpheus interview for Asha and list her weakest answers.
- Which candidates for the FDE role are still waiting on a screening response?
Automate the busywork
Talent acquisition teams
- Invite the top 10 ranked applicants to an AI interview that closes Friday 6 pm IST.
- Draft a job for a Cybersecurity Analyst in Bengaluru and save it as a draft.
- Approve the queued rejection emails for the Data Engineer role after I review them.
Build a sourcing and ranking agent
Agent builders using Claude Code, the Agent SDK, Cursor or n8n
- Search our talent pool for ML engineers in Pune with 4+ years, rank them for the open role and shortlist the top five.
- Every morning, rank new applicants for each open role and post the top three to me.
- Pair Napplied with the GitHub MCP server: check what each shortlisted engineer has actually built.
CONNECT
Works with the assistants your team already uses.
Claude (claude.ai, desktop)
Settings → Connectors → Add custom connector → paste the URL → Connect.
ChatGPT
Developer mode → create a connector with the URL and OAuth → sign in to Napplied.
Claude Code
claude mcp add --transport http napplied https://napplied.com/employer/mcp, then /mcp → Authenticate.
Cursor / VS Code
Add the URL to mcp.json; the editor opens Napplied to sign in.
Gemini CLI
gemini mcp add --transport http napplied https://napplied.com/employer/mcp, then /mcp auth napplied.
n8n, scripts, CI
Create an optional API key in Settings → MCP config and send it as a Bearer header.
BUILDING YOUR OWN AGENT?
Why not build the sourcing agent from scratch?
- A chatbot alone cannot read your applicants, interviews or rankings. MCP is how it gets them, and Napplied already exposes them as tools.
- Scraping LinkedIn from your own browser session conflicts with LinkedIn’s user agreement. Napplied works on talent you already own: applicants, imported pools and your ATS.
- A home-made screening prompt gives a score nobody can audit. Napplied ranks against the role’s criteria with a trail of what was asked, answered and evaluated.
- Interviews are the step DIY agents skip. Morpheus runs a structured first round, so the agent ranks on evidence, not just a CV.
19 TOOLS
Everything the assistant can do in Napplied.
Jobs
- list_jobs
- get_job
- create_job (draft unless you publish)
Applicants
- list_applicants
- get_applicant
- add_applicant_note
- update_applicant_status (preview, then confirm)
Interviews
- get_interview_transcripts
- invite_to_ai_interview (preview, then confirm)
- schedule_f2f_interview (preview, then confirm)
- invite_to_screening (preview, then confirm)
Ranking
- rank_applicants
- get_ranking_status
Workflows and email
- list_workflows
- list_workflow_runs
- get_workflow_run
- list_queued_emails
- approve_queued_email (preview, then confirm)
Talent pool
- search_talent
SAFE BY DEFAULT
You stay in control of every candidate touchpoint.
- Sign in with your Napplied login; each person only sees their own employer account.
- Read-only access is one click on the approval screen.
- Anything that emails a candidate returns a preview and needs your explicit yes.
- Candidate emails, phone numbers and resume links are never sent to the AI.
- Plan limits, rate limits and a daily cap on candidate emails apply; every call is in an activity log you can revoke from.
QUESTIONS
Before you connect.
Is there an MCP server for recruiting and hiring?
Yes. Napplied runs a remote MCP server at https://napplied.com/employer/mcp (beta). It works with Claude, ChatGPT, Cursor, VS Code, Gemini CLI and other MCP clients, and covers jobs, applicants, structured AI interviews, ranking, hiring workflows and the employer’s own talent pool.
Do I need an API key to connect Napplied to Claude or ChatGPT?
No. Add the server URL and sign in with your Napplied account when the assistant asks; you approve Full access or Read-only. API keys are optional, for scripts, n8n or tools that cannot sign in.
How do I build an AI sourcing agent for recruiting?
Give an agent (Claude Code, the Claude Agent SDK, Cursor or n8n) the Napplied MCP server for your talent pool, applicants, interviews and ranking, plus the GitHub MCP server for public engineering signals. Ask it to search, rank against the role’s criteria and shortlist, and keep a human approval step before anything reaches a candidate; Napplied enforces that preview for you.
Can an AI assistant email or reject candidates by itself?
Not through Napplied without you. Status changes, interview invites, screening invites and queued emails first return a preview with a signed confirmation token, and the assistant can only act after you agree. Read-only connections cannot change anything.
Is candidate data safe when using Napplied through an AI assistant?
The assistant only sees what your account can see, never candidate contact details, and every call is logged under Settings → MCP config, where you can disconnect an app or revoke a key immediately.
Does it work for hiring in India, the US and other countries?
Yes. It is the same Napplied account and data, so it works wherever your team hires, including India and the United States.
Connect Napplied in two minutes.
Add https://napplied.com/employer/mcp as a custom connector and sign in. Compare how Napplied covers search, screening, interviews and ranking on Why Napplied.