job-mcp
Enables multi-source tech job search and application with CV-based skill extraction, deduplication, and scoring across HireMeTech, Comeet, and AllJobs, supporting autonomous or supervised workflows.
README
Universal Multi-Source Job Search FastMCP Server (job-mcp)
An enterprise-grade, privacy-first FastMCP server providing intelligent, multi-source tech job aggregation, smart deduplication, dynamic CV skill extraction, requirement coverage scoring, and autonomous job application workflows across HireMeTech, Comeet ATS, and AllJobs Israel.
Architecture Overview
graph TD
Client([MCP Client: Claude / Cursor / Gemini Spark / Antigravity]) --> Tools[FastMCP Server Layer]
Tools --> Aggregator[JobAggregator]
Aggregator --> Registry[SourceRegistry]
subgraph Parallel Pluggable Sources Layer
Registry --> S1[HireMeTechSource<br/>Direct REST API + Session Fallback]
Registry --> S2[ComeetSource<br/>Direct ATS API + Concurrency Semaphore]
Registry --> S3[AllJobsSource<br/>Category Feeds + Anti-Blocking Headers]
end
subgraph Processing & Normalization Engine
S1 --> Dedup[Deduplication & Entity Merger]
S2 --> Dedup
S3 --> Dedup
Dedup --> NormKey["Key = slug(title) + '@' + slug(company)"]
NormKey --> Merge[Metadata & Links Merger]
Merge --> Scorer[Unified CV / Skill Matcher]
end
subgraph Dynamic Candidate Engine
CV["Candidate CV (.pdf / .docx / .txt)"] --> Extractor[Dynamic CV & Profile Extractor]
Extractor --> Skills["Extracted Skills (40+ tokens)"]
Extractor --> Seniority["Inferred Seniority & Exclusions"]
Extractor --> Roles["Target Job Roles"]
Skills --> Scorer
Seniority --> Scorer
end
Scorer --> Cache[Unified JobCache - 1h TTL]
Cache --> Tools
Key Features
-
Dynamic CV & Candidate Profile Extraction:
- Multi-Format Ingestion: Supports
.pdf(viapypdf),.docx(viapython-docx), and.txtfiles. - NLP Skill Chunking & Dynamic Lexicon: Discovers and extracts 40+ technical skills without brittle hardcoding, supporting complex multi-word technologies (e.g., FastAPI, LangGraph, PostgreSQL, Smart Contracts, GraphRAG).
- Automatic Seniority & Exclusion Detection: Accurately infers candidate seniority (Junior, Mid, Senior, Lead, Principal, Director) and generates intelligent negative keywords to filter out mismatched positions.
- Stopword & Noise Filtering: Rigorously discards resume structural artifacts, dates, education titles, and non-technical metadata.
- Multi-Format Ingestion: Supports
-
Smart Requirement Coverage Scoring (0–100):
- Job Requirement Coverage Ratio: Computes how comprehensively candidate skills satisfy the specific job's listed tech stack (
matched_job_skills / total_job_skills), preventing penalty for candidates with broad resumes. - Weighted Component Scoring:
- Tech Stack Overlap & Coverage: Up to 40 points
- Full CV Keyword Relevance: Up to 25 points
- Work Mode & Location Alignment: Up to 20 points
- Salary Expectations: Up to 15 points
- Exclusion Penalty: -100 points for hard seniority/tech disqualifiers.
- Tiered Match Categorization:
- Top-Tier Match ($\ge 85$): Auto-apply / prioritized application candidates.
- Strong Match ($70 - 84$): High-interest listings flagged for review/bookmarking.
- Disqualified ($< 50$): Automatically hidden or purged.
- Job Requirement Coverage Ratio: Computes how comprehensively candidate skills satisfy the specific job's listed tech stack (
-
Pluggable Multi-Source Architecture:
- HireMeTech: Direct REST API integration (
/api/jobs/search,/api/auth/me,/api/resume/profile) with automated DOM fallback. - Comeet (Direct ATS): Direct integration with Comeet Careers API (
/careers-api/2.0/company/{id}/positions) withasyncio.Semaphore(5)rate-limiting, tech directory indexing, and per-company TTL caching. - AllJobs Israel: Category feed integration with realistic browser headers and source-level error isolation.
- HireMeTech: Direct REST API integration (
-
Cross-Source Deduplication & Entity Merger:
- Eliminates duplicates when listings appear across multiple job boards.
- Merges source lists (
sources: ["hiremetech", "comeet"]), unions tech stacks, preserves richest description, and prioritizes direct ATS application links.
-
Autonomous & Supervised Operation Modes:
- Supervised Mode: Standard MCP confirmation for each tool.
- Autonomous Mode: Safe read/filter/bookmark chaining without manual prompts; two-stage safety barrier on application submission.
-
Observability & Resilience:
- Structured JSON logging (
structlog) writing tostderrwith token/credential sanitization. - Automatic trace ID tracking across all
ToolResponsepayloads.
- Structured JSON logging (
Tool Reference (9 Tools)
| Tool Name | Parameters | Description |
|---|---|---|
list_job_sources |
none | Lists all registered job sources (hiremetech, comeet, alljobs), capabilities, and real-time health. |
get_job_matches |
sources: list[str] = None, force_refresh: bool = False |
Fetches matched listings across all or specified platforms with deduplication. |
filter_jobs_by_preferences |
tech_stack: list[str], work_mode: str, location: str, min_salary: int, keywords: list[str], exclude_keywords: list[str], cv_path: str |
Scores and filters aggregated jobs against candidate CV and preferences. |
bookmark_job |
job_id: str |
Saves/favorites a job listing on the originating platform. |
delete_job |
job_id: str |
Dismisses/hides a job listing from view and removes it from cache. |
auto_apply_job |
job_id: str |
Step 1: Inspects application modal, stages preview, reports warnings. |
confirm_auto_apply |
job_id: str |
Step 2: Executes application submission. Always requires explicit confirmation. |
calibrate_selectors |
none | Discovers and calibrates DOM selectors against live pages with self-healing heuristics. |
set_operation_mode |
mode: 'supervised' | 'autonomous' |
Switches server execution mode between supervised and autonomous. |
Quick Start & Setup
1. Clone & Install Dependencies
git clone https://github.com/zvieli/hireme_mcp.git
cd hireme_mcp
# Using uv (recommended)
uv venv .venv
uv pip install -e ".[dev]"
playwright install chromium
2. Configure Your Candidate Profile & CV
Place your resume (cv.pdf, cv.docx, or cv.txt) in the root directory:
cp /path/to/your/resume.pdf ./cv.pdf
cp .env.example .env
Edit .env to configure your default CV path and contact details:
DEFAULT_CV_PATH=./cv.pdf
CANDIDATE_EMAIL=your.email@example.com
CANDIDATE_NAME="Your Name"
3. (Optional) First-Time Authentication Setup for HireMeTech
Comeet and AllJobs work automatically without login. To authenticate your HireMeTech account for direct API access and auto-apply:
.venv/bin/python -m job_mcp.setup
- A Chromium browser window will open.
- Log in with your credentials.
- Return to the terminal and press
[Enter]to save the session to./browser_profile.
Running the Server
Option A: Using Docker (Recommended)
# Build and run in background
docker compose up -d
# View live multi-source aggregation logs
docker compose logs -f hireme-mcp
Option B: Local Execution
# Streamable HTTP (Default for Web & Cloud Clients)
.venv/bin/python -m job_mcp --transport http --host 0.0.0.0 --port 8000
# Stdio (Default for Desktop Clients)
.venv/bin/python -m job_mcp --transport stdio
Visual CLI Pipeline Runner
To run a full autonomous discovery, scoring, and application dry-run directly in your terminal with rich visual output:
# Run with auto-extracted skills from your CV:
.venv/bin/python scripts/run_mock_llm_pipeline.py --cv ./cv.pdf
# Run with explicit stack override and remote filter:
.venv/bin/python scripts/run_mock_llm_pipeline.py --cv ./cv.pdf --stack "Python,FastAPI,LangGraph" --work-mode remote --location "Tel Aviv"
# Execute live application submissions (disabled by default in dry-run):
.venv/bin/python scripts/run_mock_llm_pipeline.py --cv ./cv.pdf --auto-apply
MCP Client Configuration
1. Claude Desktop (claude_desktop_config.json)
On Linux: ~/.config/Claude/claude_desktop_config.json
On macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
On Windows: %APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"job-search-mcp": {
"command": "/absolute/path/to/hireme_mcp/.venv/bin/python",
"args": ["-m", "job_mcp", "--transport", "stdio"],
"env": {
"BROWSER_HEADLESS": "true",
"DEFAULT_CV_PATH": "/absolute/path/to/hireme_mcp/cv.pdf",
"CANDIDATE_EMAIL": "candidate@example.com",
"LOG_LEVEL": "INFO"
}
}
}
}
2. Gemini Spark / Web MCP Clients
- Endpoint URL:
https://<your-host-or-devtunnel-id>/mcp - Transport:
Streamable HTTP - Authentication: None / No Auth
Environment Variables
| Variable | Default | Description |
|---|---|---|
DEFAULT_CV_PATH |
./cv.pdf |
Default CV file path for dynamic candidate skill extraction. |
CANDIDATE_EMAIL |
candidate@example.com |
Candidate email for automated application modals. |
CANDIDATE_NAME |
"" |
Candidate full name for application forms. |
MCP_TRANSPORT |
http |
Transport protocol (http, sse, stdio). |
MCP_HOST |
0.0.0.0 |
Host binding for HTTP/SSE transport. |
MCP_PORT |
8000 |
Port for HTTP/SSE transport. |
BROWSER_HEADLESS |
true |
Run browser in headless mode (true/false). |
BROWSER_PROFILE_DIR |
./browser_profile |
Directory for persistent Chromium session storage. |
CACHE_TTL_MINUTES |
60 |
In-memory deduplicated job cache TTL in minutes. |
LOG_LEVEL |
INFO |
Structured logging level (DEBUG, INFO, WARNING, ERROR). |
Running Tests
Run the full automated test suite (542 tests):
.venv/bin/pytest tests/ -v
License
This project is licensed under the MIT License.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.