HackerRank MCP
HackerRank MCP lets approved AI clients retrieve assessment, candidate, interview, team, and audit information from HackerRank. Recruiters and interviewers can review pipeline activity, prepare interview briefs, and create reports without changing HackerRank data.
This guide documents 45 customer-facing, read-only tools backed by HackerRank APIs. Configure the server at https://www.hackerrank.com/work/mcp.
What is MCP?
Model Context Protocol (MCP) is an open standard that lets AI applications work with external products. HackerRank provides the MCP server; Claude Code or ChatGPT Codex is the client where you ask questions and give instructions.
Connect
Add the production URL to your preferred MCP client.
Authenticate
Sign in with OAuth, or give an autonomous agent a protected personal API key.
Ask
Use natural language; your assistant chooses the relevant HackerRank tools.
One connection, role-specific workflows
Use the same HackerRank connection for the work that matters to your role.
Keep the pipeline visible
Review candidate progress, identify qualified candidates, track invites, and prepare weekly hiring updates.
Prepare and follow up
See upcoming interviews, retrieve interview details, and summarize transcripts and reports.
Understand assessment content
Inspect tests and questions, review available API operations, and retrieve structured data from versioned read endpoints.
Choose your MCP client
The server is live in production. Interactive clients can start OAuth sign-in automatically; autonomous agents can use a personal V3 API key as a Bearer credential.
Claude Code CLI
Add the server from your terminal, then authenticate from the /mcp menu.
Claude Code GUI
Add HackerRank as a remote connector, then enable it in your desktop Claude Code session.
ChatGPT Codex
Add the production URL in Settings, then use the Authenticate action.
Autonomous agent
Use a settings-generated personal V3 API key from a protected environment variable.
Connect from your terminal
Add the production server for your user account:
claude mcp add hackerrank \ --scope user \ --transport http \ https://www.hackerrank.com/work/mcp
claude mcp addAdd and save the HackerRank server.claude mcp listList every configured MCP server.claude mcp get hackerrankInspect HackerRank's URL, transport, and scope./mcpRun inside Claude Code to authenticate, check connection status, view the tool count, retry, or clear authentication.claude mcp remove hackerrankRemove the saved HackerRank server configuration.- Run
claudein your terminal to start Claude Code. - Inside Claude Code, run
/mcp. - Find HackerRank. If it shows needs authentication, select it and press Enter.
- Complete the HackerRank sign-in in the browser window, then return to Claude Code.
- Open
/mcpagain and confirm that HackerRank shows as connected with its available tool count.
To change the server URL or scope, remove the HackerRank server and add it again with the new settings. Advanced Claude Code commands also include claude mcp add-json, claude mcp add-from-claude-desktop, and claude mcp reset-project-choices; they are not required for this hosted user-scoped setup.
Connect from Claude Desktop
HackerRank is a hosted remote MCP server. Add it as a custom connector to make it available in Claude Desktop and your desktop Claude Code sessions.
- Open Claude Desktop and go to Customize → Connectors.
- Select +, then Add custom connector.
- Enter
https://www.hackerrank.com/work/mcpas the remote MCP server URL. Leave the optional OAuth client ID and client secret empty. - Select Add, then find HackerRank in the connector list and select Connect.
- Complete the HackerRank sign-in in your browser, then return to Claude Desktop.
- Start or open a Claude Code session. From the + menu, open Connectors and enable HackerRank for that session.
On Team and Enterprise plans, an Owner or Primary Owner must first add the connector from Organization settings → Connectors → Add → Custom → Web. Team members can then connect their own HackerRank account from Customize → Connectors.
For Claude account and organization-specific details, see Anthropic's remote custom connector guide.
Connect from ChatGPT Codex
Add the HackerRank production MCP directly in ChatGPT Codex.
- Open Settings → Plugins → MCP.
- Select Add Server and enter
https://www.hackerrank.com/work/mcp. - Return to the MCP list and select Authenticate next to HackerRank.
- Complete the HackerRank sign-in in your browser.
- Return to ChatGPT Codex and confirm that the server shows as connected.
https://www.hackerrank.com/work/mcpConnect an autonomous agent
Use this path for a non-interactive agent that cannot complete browser OAuth. The agent connects to the same production URL and sends the personal key with the Bearer scheme.
- Sign in to HackerRank for Work and open your API token settings.
- Generate a personal V3 API key with a label that identifies the agent.
- Copy the key once and store it in a secret manager or protected environment variable. Do not put it in source control or an MCP configuration file.
- Configure the remote MCP server to read that environment variable as its Bearer token.
- Initialize MCP, list tools, and call
check_api_connectionto verify access. - Revoke the key from HackerRank settings when the agent no longer needs access.
export HACKERRANK_MCP_API_KEY='<personal-v3-api-key>' codex mcp add hackerrank-agent \ --url https://www.hackerrank.com/work/mcp \ --bearer-token-env-var HACKERRANK_MCP_API_KEY
The key must belong to the current user and an MCP-enabled company. Partner, machine-to-machine, expired, revoked, unrecognized, and MCP-generated downstream credentials are rejected. Basic authentication and bare keys are not accepted.
Authentication
HackerRank MCP accepts OAuth access tokens and personal V3 API keys concurrently on POST /work/mcp. Both use Authorization: Bearer <credential>; HackerRank validates and classifies every opaque credential.
OAuth for interactive clients
- Add the production MCP URL to Claude Code or ChatGPT Codex without a custom authorization header.
- Start authentication from the MCP server list.
- Sign in from your browser, then return to your client.
- Confirm that HackerRank shows as connected before asking it to use HackerRank.
Personal key for autonomous agents
Generate the key from your HackerRank token settings and keep it in a secret manager. The key uses your current HackerRank identity and company eligibility, and it can be revoked independently at any time.
The current HackerRank MCP is read-only. Its dedicated tools and raw API request tool retrieve data through GET requests; they do not create, update, archive, invite, or delete HackerRank data.
Prompts to try
Copy any prompt below and edit it as needed.
“Create a weekly recruiting update from HackerRank for my active tests. Include invitations, attempts, completions, qualified candidates, and next actions.”
“Show my upcoming HackerRank interviews this week and prepare a short briefing for each candidate.”
“Summarize this test’s candidate pipeline in HackerRank by assessment status and ATS stage. Highlight candidates who need follow-up.”
“In HackerRank, list the questions attached to this test and summarize the available details for each coding question.”
“Review yesterday’s HackerRank interview transcript and report. Summarize strengths, concerns, and follow-up topics.”
“Use HackerRank test activity and candidate outcomes to create a structured summary for a recruiting dashboard.”
Capabilities reference
All 45 read-only tools shown on this page, grouped by capability. This includes candidate, assessment, interview, team, question, invite template, and audit workflows.
Users
3 toolslist_usersget_usersearch_usersTeams
4 toolslist_teamsget_teamlist_team_membersget_team_membershipTests
5 toolslist_testssearch_testsget_testlist_test_invitersget_test_questionsCandidates & pipeline
16 toolssearch_candidateslist_candidatessearch_test_candidatesget_candidate_resultget_candidate_report_pdfget_candidate_by_emailfilter_candidates_by_statusfilter_candidates_by_ats_stateget_qualified_candidateslist_test_attemptsrank_test_candidatesfind_candidates_needing_reviewsummarize_test_pipelinefind_stale_invitesget_test_summaryget_pipeline_overviewInterviews & preparation
9 toolslist_interviewsget_interviewget_interview_transcriptget_interview_reportlist_upcoming_interviewsfind_unscheduled_codepair_candidatessummarize_candidate_reportcompare_candidate_resultsprepare_interview_packetInterview templates
2 toolslist_interview_templatesget_interview_templateQuestions
3 toolslist_questionssearch_questionsget_questionInvite templates & audit
3 toolslist_invite_templatesget_invite_templatelist_audit_logsAuthentication observability
Every authentication attempt emits one secret-safe mcp_auth_result log. Facet it by credential mode, outcome, and sanitized reason:
FROM Log SELECT count(*) WHERE service = 'hackerrank-mcp' AND message = 'mcp_auth_result' FACET auth_method, auth_outcome, reason
For an introspection cache miss, use the MCP event's auth_correlation_id to find HRW's mcp_credential_introspection event with the same request_id.
FROM Transaction SELECT count(*) WHERE oauthEvent = 'mcp_credential_introspection' FACET oauthAuthMethod, oauthAuthOutcome, oauthReason
Logs never include OAuth tokens, API keys, authorization headers, introspection bodies, MCP request bodies, tool arguments, downstream credentials, or tool response content. The salted one-way credential_subject is the only credential identifier.
Learn with step-by-step videos
A video series will walk through practical HackerRank MCP workflows for recruiters, interviewers, and developers.
Automate weekly hiring updates
Turn assessment and candidate activity into a concise recurring update.
Prepare for upcoming interviews
Gather schedules, candidate context, transcripts, and reports.
Create recruiting dashboards
Shape test and pipeline data into decision-ready dashboard inputs.