The Greenhouse MCP server connects Greenhouse Recruiting to Claude. Anthropic's directory lists 159 tools for candidates, applications, jobs, interviews, scorecards, offers, users and reporting, about half of them writes. It can move, reject or hire an application. Every tool sits behind scopes a Site Admin enables and each user's own Greenhouse permissions.
Anthropic states this reflects the level of review a connector received, not a security audit.
Greenhouse tools (159)
activate_user
add_candidate_note
anonymize_candidate
bulk_activate_users
bulk_create_blocked_spam_sources
bulk_create_candidates
bulk_create_custom_field_options
bulk_create_jobs
bulk_create_openings
bulk_create_users
bulk_deactivate_users
bulk_revoke_permissions_users
bulk_update_blocked_spam_sources
bulk_update_custom_field_options
bulk_update_departments
bulk_update_offices
bulk_update_openings
bulk_update_users
convert_prospect_to_candidate
create_application
create_applied_candidate_tag
create_approval_flow
create_attachment
create_blocked_spam_source
create_candidate
create_candidate_education
create_candidate_employment
create_candidate_tag
create_custom_field
create_custom_field_department
create_custom_field_office
create_custom_field_option
create_department
create_future_job_permission
create_interview
create_job
create_job_hiring_manager
create_job_note
create_job_owner
create_job_post
create_job_post_location
create_offer
create_office
create_opening
create_scorecard
create_scorecard_candidate_attribute
create_scorecard_question_answer
create_scorecard_question_answer_option
create_user
create_user_email
create_user_job_permission
deactivate_user
get_bulk_request
get_job
get_rejection_reason
hire_application
list_ai_opt_out_statuses
list_application_stages
list_applications
list_applied_candidate_tags
list_approval_flows
list_approver_groups
list_approvers
list_attachments
list_blocked_spam_sources
list_bulk_requests
list_candidate_attribute_types
list_candidate_educations
list_candidate_employments
list_candidate_tags
list_candidates
list_close_reasons
list_custom_field_departments
list_custom_field_offices
list_custom_field_options
list_custom_fields
list_default_interviewers
list_demographic_answer_options
list_demographic_answers
list_demographic_question_sets
list_demographic_questions
list_departments
list_eeoc
list_email_templates
list_focus_candidate_attributes
list_future_job_permissions
list_interview_kits
list_interview_media
list_interviewer_tags
list_interviewers
list_interviews
list_job_board_custom_locations
list_job_boards
list_job_candidate_attributes
list_job_hiring_managers
list_job_interviews
list_job_notes
list_job_owners
list_job_post_locations
list_job_post_searchable_locations
list_job_posts
list_job_stages
list_jobs
list_notes
list_offers
list_offices
list_openings
list_pay_input_ranges
list_pay_inputs
list_prospect_details
list_prospect_pool_stages
list_prospect_pools
list_referrers
list_rejection_details
list_rejection_reasons
list_scorecard_candidate_attributes
list_scorecard_question_answer_options
list_scorecard_question_answers
list_scorecard_question_candidate_attributes
list_scorecard_question_options
list_scorecard_questions
list_scorecards
list_sources
list_tracking_links
list_user_emails
list_user_job_permissions
list_user_roles
list_users
merge_candidate
move_application
reject_application
replace_approver_approver_group
replace_approver_groups_approval_flow
request_approvals_approval_flow
revoke_user_permissions
search_applications
search_candidates
search_jobs
unreject_application
update_application
update_approval_flow
update_blocked_spam_source
update_candidate
update_custom_field
update_custom_field_option
update_department
update_interview
update_job
update_job_note
update_job_post
update_offer
update_office
update_opening
update_rejection_detail
update_scorecard
update_scorecard_candidate_attribute
update_scorecard_question_answer
update_user
whoami
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
Limits
Open beta. Greenhouse says capabilities will keep expanding, so the tool list can change.
Tier-gated. Core, Plus and Pro only; legacy tiers need to talk to their account team.
Every client needs an approved redirect URL. A tool that is not added under MCP Access → Clients cannot connect, and removing a redirect URL disconnects everyone using it.
Missing tools usually mean missing scopes. Tools for disabled scopes do not appear.
Data quality drives answer quality. Greenhouse advises consistent stage names, complete scorecards and structured custom fields for better analysis.
Candidate notifications are unconfirmed. We could not establish whether rejecting an application or scheduling an interview through the connector emails anyone.
We could not read tool schemas or annotations. The endpoint returned 401 to an anonymous request. Tool names come from Anthropic's directory; the read/write split is by name.
Frequently asked questions
Can Greenhouse MCP take actions, or is it read-only?
It can take actions. Greenhouse states the MCP supports both read and write actions and is not limited to read-only access. Whether a given action is available depends on what Greenhouse MCP supports today, which scopes your organization enabled in Dev Center under MCP Access, and your own Greenhouse permissions. Until an admin saves that page, connections get seven read-only list scopes.
Does Greenhouse MCP follow my Greenhouse permissions?
Yes. Greenhouse says users cannot reach data or take actions beyond what their existing Greenhouse permissions and the organization's approved scopes allow. The scopes a Site Admin enables in Dev Center under MCP Access act as a ceiling, so a user cannot grant themselves more at sign-in. A recruiter and a Site Admin connecting the same AI tool therefore see different capabilities.
What happens if I approve the wrong Greenhouse scopes?
Disconnect and reconnect. Greenhouse says to disconnect the connector in your AI tool, reconnect, and complete authorization with the correct scopes. Newly added scopes are not applied automatically, so users must reconnect to pick them up. If the organization adds or removes scopes, Greenhouse signs all users out, which means every connected user has to authorize again.
Does my AI tool get a client secret from Greenhouse?
No. Greenhouse always registers an AI tool as a public client. The registration response sets the token endpoint authentication method to none and carries no client secret, even if the tool asked for one, and each sign-in is proven with PKCE instead. A custom client or proxy that waits for a secret cannot complete the connection, per Greenhouse.
Who can use Greenhouse MCP?
Users on the GHR Core, Plus and Pro subscription tiers. Greenhouse says the MCP is in open beta and not yet generally available. A Site Admin completes the first setup and chooses the scopes, then other users on eligible tiers, such as Job Admins and recruiters, connect their own AI tools. Organizations on a legacy tier should ask their account team.
Live OAuth posture check: an anonymous initialize to https://mcp.greenhouse.io/mcp returned 401 with a bearer authentication challenge naming https://mcp.greenhouse.io/.well-known/oauth-protected-resource/mcp, which lists https://auth.greenhouse.io and the single scope greenhouse:mcp_client. The vendor's Claude Code host https://mcp.us.greenhouse.io/mcp returned the same kind of challenge (2026-11-18) · retrieved 2026-11-18
Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.
Use it in an agent
Put Greenhouse to work.
Connect Greenhouse once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.