Ai Hr Management Toolkit

io.github.XJTLUmedia由 GitHub 账号 XJTLUmedia 发布v3.0.5更新于 Oct 24, 2026

AI HR toolkit: 24 MCP tools for resume parsing, skill extraction & ATS management.

概览

AI 生成的概览

一个本地 MCP 服务器,让 AI 助手解析简历、提取技能并给候选人打分,还提供覆盖职位、候选人、面试和录用通知的完整招聘管理系统。

可以做什么
解析简历与文档:读取 PDF、DOCX、TXT、Markdown 和网址内容,输出结构化 JSON,支持单个与批量处理,并可查看分析流程各阶段与置信度。分析候选人:提取关键词、模式、实体、技能、经历,以及与职位描述的相似度。评分与排序:按加权标准评估候选人,给出通过、复核或拒绝的结论。管理招聘流程:创建和编辑职位、候选人、面试、录用通知与备注,并提供搜索、分析、评分卡、人才库、入职清单与合规报告。导出与发送:将结果导出为 JSON 或 CSV 并通过邮件发送。生成测试用示例数据。
适用场景
适合需要让助手批量筛选简历、按职位描述比较候选人,或在不另装招聘系统的情况下维护招聘流程的场景。适合在桌面 MCP 客户端中使用的招聘人员和中小型 HR 团队。
运行要求
以 npm 包形式通过 stdio 在本地运行,通常用 npx 启动,因此需要 Node.js,且工作目录中要有 package.json。算法类工具不需要 API 密钥或环境变量;如需 LLM 辅助评分,可在单次调用中提供 AI 服务商密钥。README 还介绍了自行部署的 HTTP 端点方式。
安装前请注意
招聘管理工具会创建、更新、移动和删除职位、候选人、面试、录用通知和备注等记录,README 说明这些数据保存在内存中,可能不会持久化。邮件工具通过单次调用传入的 SMTP 凭据发送邮件,可选的 LLM 评分路径也需在调用时传入 API 密钥,请避免这些密钥出现在共享日志中。简历文件和网址会交给服务器进程处理,输入的网址会被联网抓取。

安装

在 SourceWeft 中

  1. 打开 控制台中的 Ai Hr Management Toolkit,将其添加到工作区。
  2. 为需要使用其工具的对话启用该服务。

Desktop only,通过 STDIO。 STDIO 服务会启动本地进程,因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照 仓库 中的启动说明。

README

AI HR Management Toolkit

AI-powered resume parser & full Applicant Tracking System with 21 MCP tools. Parse PDFs, extract skills, detect patterns, score candidates, and manage a complete hiring pipeline — all from your AI assistant, no manual work required.

[image] [image] [image]

Live demo: https://ai-hr-management-toolkit.vercel.app

[npm version] [License: MIT]

[mcp-ai-hr-management-toolkit server]

What Is This?

You have 50 resumes to screen. Your AI assistant can reason about candidates — but it cannot open PDFs, extract structured data, or track pipeline stages. This toolkit bridges that gap.

Give your AI assistant 21 tools covering the entire hiring workflow:

  • Parse PDFs, DOCX, TXT, Markdown, and URLs into structured JSON
  • Extract skills, experience, keywords, and entities algorithmically
  • Score and rank candidates against job descriptions
  • Run a full ATS: jobs, candidates, interviews, offers, notes, and analytics

20 of 21 tools are 100% algorithmic — no LLM calls, no API keys required. The AI calls tools, interprets the results, and delivers analysis. You just ask questions.


Quick Start (MCP Clients)

No installation needed. Point your MCP client at the package:

Claude Desktop — Edit %APPDATA%\Claude\claude_desktop_config.json (Windows) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

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

Example usage:

[image] [image]

Cursor — Add to .cursor/mcp.json in your project root:

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

VS Code Copilot — Create .vscode/mcp.json in your project root:

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

VS Code users: Run the npx command from a directory that contains a package.json (i.e. any project root). The cwd key in .vscode/mcp.json can override the working directory if needed.

Windsurf / other MCP clients — Use the same npx pattern above.


Installation Options

Option 1: NPX (Zero-install, recommended)

Works from any project directory (requires a package.json in the working directory):

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

Option 2: Global install

Install once, use from any directory:

bash
npm install -g mcp-ai-hr-management-toolkit
json
{  "mcpServers": {    "ai-hr-management-toolkit": {      "command": "mcp-ai-hr-management-toolkit",      "args": []    }  }}

Option 3: Remote HTTP endpoint

Deploy the Next.js app and use the Streamable HTTP transport:

https://your-domain.com/api/mcp

Test locally:

bash
npx @modelcontextprotocol/inspector http://localhost:3000/api/mcp

Option 4: Local development (Web UI + MCP)

bash
git clone <repo-url>cd Resume-parsernpm installnpm run dev

Web UI at http://localhost:3000. MCP endpoint at http://localhost:3000/api/mcp. No .env needed — configure API keys in the UI or pass them per tool call.


All 21 MCP Tools

All tools return structured JSON with next_steps hints so the AI knows what to call next.

Resume Parsing & Ingestion

ToolWhat it doesAI?
parse_resumeParse PDF / DOCX / TXT / MD / URL → raw text + contacts, keywords, section mapNo
batch_parse_resumesParse up to 20 files in one call, full pipeline on eachNo
inspect_pipelineRun the 5-stage analysis pipeline → confidence scores, entity counts, data quality reportNo

Unified Analysis

ToolWhat it doesAI?
analyze_resumeMaster analysis tool with selectable aspects: keywords (TF-IDF + bigrams), patterns (date ranges, metrics, team sizes, career trajectory), entities (NER with 12 types + context disambiguation), skills (13 categories with proficiency estimation), experience (structured timeline), similarity (cosine, Jaccard, TF-IDF overlap vs. job description), or allNo

analyze_resume consolidates what were previously 7 separate tools (extract_keywords, detect_patterns, classify_entities, extract_skills_structured, extract_experience_structured, compute_similarity, analyze_resume_comprehensive) into a single entry point with aspect selection.

Candidate Matching & Scoring

ToolWhat it doesAI?
assess_candidateScore against up to 8 weighted criteria axes → weighted total + pass / review / reject decisionOptional

Export & Notifications

ToolWhat it doesAI?
export_resultsExport structured parse results to JSON or CSVNo
send_emailSend results via SMTP (config passed per call — no server-side secrets stored)No

ATS — Jobs

ToolWhat it doesAI?
ats_manage_jobsFull CRUD for job postings: create, read, update, delete, list, search by title/department/statusNo

ATS — Candidates & Pipeline

ToolWhat it doesAI?
ats_manage_candidatesCRUD + analytics: add, update, move stage, bulk-move, filter, rank, compare, recommend stage changes, summarizeNo
ats_analyticsUnified dashboard + pipeline analytics: stage distribution, conversion rates, avg time-in-stage, bottleneck detection, offer acceptance rateNo
ats_searchGlobal full-text search across all ATS entities (candidates, jobs, interviews, offers, notes)No

ATS — Interviews

ToolWhat it doesAI?
ats_schedule_interviewCreate, update, and delete interviews with conflict detection and interviewer availability checkNo
ats_interview_feedbackSubmit structured feedback, compute consensus score, summarize feedback across all interviewersNo

ATS — Offers & Notes

ToolWhat it doesAI?
ats_manage_offersFull offer lifecycle: draft → pending → approved → sent → accepted / declined / expiredNo
ats_manage_notesAdd, update, search, and delete timestamped candidate notesNo

ATS — Enterprise HR

ToolWhat it doesAI?
ats_complianceEEO/EEOC reporting, GDPR export/erasure, audit trail, data retention policiesNo
ats_talent_poolPassive candidate talent pools (CRM): create pools, add/remove candidates, search, analyticsNo
ats_scorecardStructured interview scorecards with weighted criteria, per-evaluator scores, aggregate rankingsNo
ats_onboardingPost-hire onboarding checklists: tasks by category, assignees, progress tracking, overdue alertsNo
ats_communicationEmail templates with {{variable}} interpolation, send/preview, communication history, statsNo

Testing & Seeding

ToolWhat it doesAI?
ats_generate_demo_dataGenerate a realistic sample ATS dataset (jobs, candidates, interviews, offers) for testingNo

assess_candidate optionally calls an LLM when you supply provider + apiKey; it falls back to fully algorithmic scoring otherwise.


Example Multi-Turn Flow

You: "Parse this resume and tell me if they're a good fit for our Senior Engineer role"
AI → parse_resume(file)     → raw text, contact info, section map
AI → inspect_pipeline(rawText)     → 5-stage confidence scores, entity classification
AI → analyze_resume(text, aspects=["skills", "patterns", "similarity"], jobDescription=...)     → 13 skill categories with proficiency levels     → career trajectory, metrics, date ranges     → cosine 0.74, skill match 82%, gap analysis
AI synthesizes → "Strong match. 6 of 8 required skills present.                  Two gaps: Kubernetes and system design at scale.                  Recommend: Technical Screen"

Analysis Pipeline

Every resume runs through a 5-stage algorithmic pipeline:

┌─────────────┐    ┌──────────────┐    ┌──────────────┐    ┌────────────────┐    ┌───────────────┐│  Ingestion  │───▶│ Sanitization │───▶│ Tokenization │───▶│ Classification │───▶│ Serialization ││ (file/URL)  │    │ (noise trim) │    │  (TF-IDF)    │    │ (NER + disamb) │    │ (structured)  │└─────────────┘    └──────────────┘    └──────────────┘    └────────────────┘    └───────────────┘
  1. Ingestion — PDF via pdf-parse v2, DOCX via mammoth, HTML/URL via cheerio, plain text/markdown natively
  2. Sanitization — Removes non-ASCII artifacts, normalizes whitespace, strips formatting noise
  3. Tokenization — TF-IDF with unigrams, bigrams, and trigrams; scored by document frequency
  4. Classification — NER with domain-aware disambiguation (e.g. "Java" as language vs. Indonesian city; "Go" as language vs. verb)
  5. Serialization — Maps entities to typed ResumeSchema with confidence scores and data quality metrics

Supported File Formats

FormatExtensionsParser
PDF.pdfpdf-parse v2
DOCX.docxmammoth
Plain text.txtdirect read
Markdown.md, .markdownregex-based
URL / HTMLany URL stringcheerio

Max file size: 10 MB


Structured Output Schema

contact        — name, email, phone, location, LinkedIn, GitHub, website, portfoliosummary        — professional summary textskills[]       — name, category (13 types), proficiency, usage contextexperience[]   — company, title, start/end dates, highlights, achievements (with metrics), technologieseducation[]    — institution, degree, field, dates, GPAcertifications[] — name, issuer, date, credential URLprojects[]     — name, description, URL, technologies, highlightslanguages[]    — spoken language and proficiency

Web UI

The app ships with a full web interface:

TabDescription
Single ParseUpload one file or paste a URL. Returns structured data, pipeline visualization, and AI-enhanced analysis
Batch ParseUpload up to 20 files. Export to JSON / CSV / PDF or email results
ChatConversational interface with tool access — ask questions about any parsed resume
ATSFull pipeline board: jobs, candidates (Kanban), interviews, offers, and analytics dashboard

Switch AI providers from the selector at the top. Supports OpenAI, Anthropic, Google, DeepSeek, GLM, Qwen, OpenRouter, and OpenCode Zen.


REST API Endpoints

All endpoints accept multipart/form-data with optional headers:

HeaderDescription
x-api-keyYour AI provider API key
x-ai-provideropenai / anthropic / google / deepseek / glm / qwen / openrouter / opencodezen
x-ai-modelSpecific model ID
bash
# Parse a single resumecurl -X POST http://localhost:3000/api/parse \  -H "x-api-key: sk-..." \  -F "[email protected]"
# Batch parse (up to 20 files)curl -X POST http://localhost:3000/api/batch-parse \  -H "x-api-key: sk-..." \  -F "[email protected]" \  -F "[email protected]"
# MCP endpoint (Streamable HTTP)curl -X POST http://localhost:3000/api/mcp \  -H "Content-Type: application/json" \  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'
# Export parsed datacurl -X POST http://localhost:3000/api/export \  -H "Content-Type: application/json" \  -d '{"format":"csv","results":[...]}'

Tech Stack

LayerTechnologies
FrameworkNext.js 16 (App Router, Turbopack), React 19, TypeScript
AIVercel AI SDK v6, multi-provider (OpenAI, Anthropic, Google, DeepSeek, GLM, Qwen, OpenRouter)
MCP@modelcontextprotocol/sdk v1.29 — Streamable HTTP + stdio transports
Parsingpdf-parse v2, mammoth, cheerio
NLPTF-IDF, NER, cosine similarity, Jaccard index (all in-process, no external services)
SchemaZod v4
ExportExcelJS (CSV/XLSX), jsPDF + jspdf-autotable
EmailNodemailer
StylingTailwind CSS v4, Framer Motion

Development

bash
npm install
# Start dev server (Web UI at :3000 + MCP at /api/mcp)npm run dev
# Build the standalone MCP CLI (stdio transport)npm run build:mcp
# Build the Next.js app for productionnpm run build
# Test MCP with the official inspectornpx @modelcontextprotocol/inspector http://localhost:3000/api/mcpnpx @modelcontextprotocol/inspector node dist/mcp-stdio.js
# Lintnpm run lint

Project Structure

src/├── app/│   ├── page.tsx              # Main UI (tabs, provider selector, chat, ATS)│   ├── layout.tsx            # Root layout + global styles│   └── api/│       ├── parse/route.ts    # Single resume parse│       ├── batch-parse/route.ts│       ├── chat/route.ts     # Conversational AI with tool access│       ├── mcp/route.ts      # MCP server (Streamable HTTP)│       ├── models/route.ts   # Provider model listing│       ├── export/route.ts   # JSON / CSV / PDF export│       └── email/route.ts    # SMTP email├── components/               # React UI components (parse, batch, chat, ATS)│   └── ats/                  # ATS-specific views (Kanban, Dashboard, Scheduler…)└── lib/    ├── ai-model.ts           # Multi-provider model config (no env fallback)    ├── mcp-server.ts         # MCP server — registers all 21 tools    ├── schemas/    │   ├── resume.ts         # Zod v4 ResumeSchema    │   └── criteria.ts       # Assessment criteria schema    ├── analysis/    │   ├── pipeline.ts       # 5-stage pipeline orchestrator    │   ├── sanitizer.ts      # Text cleaning    │   ├── keyword-extractor.ts  # TF-IDF    │   ├── classifier.ts     # NER with context disambiguation    │   ├── pattern-matcher.ts    # Regex extraction (metrics, dates, contacts)    │   └── scoring.ts        # Cosine similarity, Jaccard, skill matching    ├── parser/    │   ├── pdf.ts, docx.ts, text.ts, markdown.ts, url.ts    │   └── index.ts    ├── ats/    │   ├── types.ts          # ATS entity types    │   ├── store.ts          # In-memory ATS state    │   ├── demo-data.ts      # Realistic seed data generator    │   └── context.tsx       # React context for ATS state    └── tools/        ├── parse-resume.ts       # parse_resume        ├── inspect-pipeline.ts   # inspect_pipeline        ├── export-results.ts     # export_results        ├── send-email.ts         # send_email        └── mcp/                  # 17 MCP-specific tools            ├── analyze-resume.ts     # analyze_resume (unified: keywords, patterns, entities, skills, experience, similarity)            ├── batch-parse.ts        # batch_parse_resumes            ├── assess-candidate.ts   # assess_candidate            ├── ats-manage-candidates.ts  # ats_manage_candidates (includes rank/filter/compare/summarize)            ├── ats-manage-jobs.ts            ├── ats-manage-offers.ts            ├── ats-manage-notes.ts            ├── ats-analytics.ts      # ats_analytics (unified dashboard + pipeline)            ├── ats-schedule-interview.ts            ├── ats-interview-feedback.ts            ├── ats-search.ts            ├── ats-generate-demo-data.ts            ├── ats-compliance.ts     # Enterprise: EEO / GDPR / audit            ├── ats-talent-pool.ts    # Enterprise: passive candidate CRM            ├── ats-scorecard.ts      # Enterprise: structured scorecards            ├── ats-onboarding.ts     # Enterprise: onboarding checklists            └── ats-communication.ts  # Enterprise: email templates & history

License

MIT

来源:README.md,提交 cbeba6c

工具

0
SourceWeft 尚未在沙箱中启动过该软件包,因此其工具未知。

版本历史

1
  1. v3.0.5最新Sep 16, 2026

评分

已满足适用检查项的 50%

该评分衡量此收录条目的完整程度,依据注册中心条目、端点与公开来源计算得出。它不是对服务器本身的评价,也不表示服务器安全或当前可用:可连通性是带日期的检查结果,评分也不包含安装量或用户评价。

  • 近 30 天内可连通25 分

    请保持端点能够响应 MCP 握手;SourceWeft 每 30 天探测一次。

  • 已声明认证方式10 分

    请在注册中心条目中声明凭据请求头,或保持端点能够响应探测,以便判定其认证方式。

  • 已列出工具15 分

    请让服务器在无凭据时响应 tools/list,或发布可在沙箱中无凭据启动的软件包。

  • 描述必需15 分
  • 图标10 分
  • 文档15 分
  • 许可证10 分