AI HR Management Toolkit

What XJTLUmedia/AI-HR-Management-Toolkit MCP server does

XJTLUmedia/AI-HR-Management-Toolkit MCP server combines document analysis with an applicant tracking workflow. Resume inputs can be read from PDF, DOCX, TXT, Markdown, or URLs and converted into structured results containing text, contact details, keywords, and section information. Further analysis can identify skills, entities, date ranges, metrics, team sizes, career progression, experience timelines, and similarity to a job description.

The ATS portion covers job postings, candidates, interviews, feedback, offers, notes, talent pools, scorecards, onboarding, communications, compliance, search, and analytics. Results are returned as structured JSON, and tools include next_steps hints to indicate possible follow-up operations.

How it works

The MCP client starts the package locally with npx, then sends tool calls to the server. A typical workflow begins with parse_resume, followed by inspect_pipeline and analyze_resume with selected analysis aspects. Candidate matching can use assess_candidate, which scores up to eight weighted criteria and produces a pass, review, or reject result.

Most of the 21 tools use local algorithmic processing rather than an external model. The exception is optional LLM-assisted assessment: supplying a provider and API key enables that path, while algorithmic scoring remains available without them. The ATS tools maintain their workflow data within the running application context and expose operations for creating, updating, searching, comparing, ranking, and summarizing records.

Setup and configuration

XJTLUmedia/AI-HR-Management-Toolkit MCP server can be configured as a stdio server with the npm package:

{
  "mcpServers": {
    "ai-hr-management-toolkit": {
      "command": "npx",
      "args": ["-y", "mcp-ai-hr-management-toolkit"]
    }
  }
}

The README documents configurations for Claude Desktop, Cursor, VS Code Copilot, Windsurf, and other MCP clients. A global npm installation is also supported. For local development, clone the repository, install dependencies, and run the development command; the web interface is served at port 3000 and the MCP endpoint is available at /api/mcp. No .env file is required. API keys can instead be configured in the UI or passed with an individual tool call.

The project also supports deployment as a Streamable HTTP endpoint through the Next.js application. The documented endpoint path is /api/mcp, but the deployed domain must be supplied by the operator.

Tools and capabilities

• Parse one resume or batch-process up to 20 files.
• Inspect parsing quality through a five-stage pipeline.
• Analyze keywords, patterns, entities, skills, experience, and job-description similarity.
• Score candidates with weighted criteria and optional LLM assistance.
• Export results as JSON or CSV and send results through SMTP settings supplied per call.
• Manage ATS records for jobs, candidates, interviews, feedback, offers, notes, talent pools, scorecards, onboarding, and communications.
• Generate demo ATS data for testing.

Limitations and notes

The toolkit does not require server-side secrets for its standard algorithmic tools. SMTP configuration is supplied when sending email, and the optional LLM assessment path requires the caller to provide a provider and API key. The README does not identify a persistent external database or a built-in authentication system, so deployments should evaluate storage and access control requirements separately.

XJTLUmedia/AI-HR-Management-Toolkit MCP server is licensed under MIT. Its compliance tools include EEO/EEOC reporting, GDPR export and erasure, audit trails, and retention policies, but those features do not by themselves establish legal compliance for a particular organization or jurisdiction.