@hirenewtalentai/mcp-server

An MCP (Model Context Protocol) server that wraps the core HireNewTalent public API hiring workflow. Drop it into Claude Desktop, Cursor, Claude Code, Cline, or any MCP-capable agent to search talent, run AI interviews, get pricing estimates, and manage engagements as native tool calls. Messaging, events, and full profile fetches remain REST-only today.

Quick install

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "hirenewtalent": {
      "command": "npx",
      "args": ["-y", "@hirenewtalentai/mcp-server"],
      "env": {
        "HIRENEWTALENT_API_KEY": "hnt_test_..."
      }
    }
  }
}

Restart Claude Desktop. The tools appear automatically.

Cursor

Open Cursor Settings → MCP and add:

{
  "hirenewtalent": {
    "command": "npx",
    "args": ["-y", "@hirenewtalentai/mcp-server"],
    "env": {
      "HIRENEWTALENT_API_KEY": "hnt_test_..."
    }
  }
}

Claude Code

claude mcp add hirenewtalent -- npx -y @hirenewtalentai/mcp-server
# Then set the env var:
export HIRENEWTALENT_API_KEY=hnt_test_...

Or add to your project's .claude/mcp.json:

{
  "hirenewtalent": {
    "command": "npx",
    "args": ["-y", "@hirenewtalentai/mcp-server"],
    "env": {
      "HIRENEWTALENT_API_KEY": "hnt_test_..."
    }
  }
}

Cline

In Cline's MCP Servers settings panel, add a new server:

• Command: npx
• Args: -y @hirenewtalentai/mcp-server
• Env: HIRENEWTALENT_API_KEY=hnt_test_...

Available tools

| Tool | Description | |------|-------------| | search_talent | Natural-language talent search. Returns ranked VA profiles and an estimated monthly cost range. | | get_interview_history | Marketplace-wide AI interview history for a talent (rubric_scores, section_breakdown, public_summary). | | get_pricing_estimate | Role-level price range (entry_standard ↔ senior_expert) without tying to a specific talent. | | generate_custom_interview | LLM-generated interview prompt + rubric for any custom role. See latency note below. | | create_engagement | Create an engagement and send an AI interview invite to the talent. | | get_engagement | Get a single engagement with score, billing, invite status, next_action, and valid_actions. | | list_engagements | List all engagements for the authenticated client. | | send_offer | Send the engine-stamped offer to the talent (no rate fields accepted — engine controls rates). | | pass_on_candidate | Pass on a candidate from the decision stage. Terminal. | | request_live_interview | Request a live (human-to-human) interview from the decision stage. | | terminate_engagement | Terminate an active engagement. Requires a reason. Terminal. | | request_replacement | Request a replacement VA for an active engagement (stays active and notifies support in live mode). | | update_engagement | Update the start_date on an engagement. Set it before the offer is accepted — acceptance anchors billing. | | list_events | Poll the append-only event feed of engagement lifecycle changes across all your engagements, since a cursor. | | get_talent | Fetch a single talent's full public profile (skills, experience, availability) by id. 404 if missing. | | list_conversations | List your message threads, newest activity first. Requires the messages scope. | | create_conversation | Open (get-or-create) a conversation with a talent. Requires the messages scope. | | get_messages | Fetch messages in a conversation, newest first, paginated. Requires the messages scope. | | send_message | Send a message in a conversation (max 5000 chars). Requires the messages scope. | | test_complete_interview | Test keys only — simulate the talent finishing the interview (invited → pending_decision). | | test_accept_offer | Test keys only — simulate the talent accepting the offer (offer_sent → active). | | test_decline_offer | Test keys only — simulate the talent declining the offer (offer_sent → passed). |

Environment variables

| Variable | Required | Description | |----------|----------|-------------| | HIRENEWTALENT_API_KEY | Yes | Your HireNewTalent API key (hnt_test_... for development, hnt_live_... for production). Also accepted as HNT_API_KEY. | | HIRENEWTALENT_API_URL | No | Override the base URL. Defaults to https://api.hirenewtalent.ai. |

Latency notes

• generate_custom_interview calls a synchronous LLM on the server. Typical response time is 20–45 seconds. The tool sets a 60-second timeout automatically. Ensure your MCP client does not time out requests shorter than 60 seconds.

Pricing is engine-controlled

talent_wage_rate on create_engagement and rate fields on send_offer are silently ignored by the API — the pricing engine v9 controls all rates. See the API documentation for details.

License

MIT — see the LICENSE file bundled with this package.

API documentation: https://hirenewtalent.ai/api-documentation