StackCheck：本地优先的技术栈情报与薪资基准工具

Tech Stack Market Intelligence Engine and Data Analytics Dashboard
_{Track technology demand, skill pairings, and compensation benchmarks from live job posts.}

Quick installation

Install the standalone binary without external dependencies:

macOS and Linux

curl -fsSL https://raw.githubusercontent.com/haydermuhib/StackCheck/main/install.sh | bash

Windows (PowerShell)

curl.exe -L "https://github.com/haydermuhib/StackCheck/releases/latest/download/StackCheck-windows-x64.exe" -o StackCheck.exe; .\StackCheck.exe

You can also download StackCheck-windows-x64.exe directly from GitHub Releases. If Windows SmartScreen prompts on first launch, select More info, then Run anyway.

Run the web dashboard:

stackcheck

Overview

StackCheck collects live job listings from HiringCafe, extracts technical requirements with section-based weighting, and visualizes market statistics in a local Streamlit dashboard. It helps engineers and analysts evaluate technology demand, inspect skill combinations, and benchmark compensation across regional job markets.

Web dashboard interface

Executive market overview	Geographic compensation distribution
Key metric cards, weighted tech skills frequency ranking, and domain category breakdown.	Country-level compensation strip scatter plot with horizontal median benchmarks.
Experience and employer analytics	Interactive job explorer
Seniority skill differentials, compensation tier table, hiring employers, and stack complexity.	Searchable posting records with parsed salaries, normalized tech stacks, and direct apply links.

Core capabilities

Comparison with standard job boards

Standard job boards return search cards, but do not calculate aggregate market metrics. StackCheck processes job descriptions into structured analytics:

Scraping design and limits

StackCheck is built to query origin servers responsibly without causing service disruption:

• Batch data retrieval: StackCheck reads server-rendered payloads containing 60 to 90 complete listings per HTTP request. Gathering 2,000 listings requires roughly 25 to 30 requests.
• Paced requests: Requests run sequentially with built-in courtesy pauses (0.35 seconds between calls) over a persistent connection.
• Local computation: Ingested listings save directly to a local SQLite database (~/.stackcheck/stackcheck.db). Skill parsing, salary calculations, and chart generation run entirely on your local machine.
• Relevance filtering: Tokenized validation checks role titles to keep unrelated jobs from polluting specific searches.

System architecture

┌──────────────────────┐     ┌──────────────────────┐     ┌──────────────────────┐
│  HiringCafe Public   │────▶│    Deduplicator and   │────▶│    Section-Aware     │
│      API Client      │     │  Country Normalizer  │     │   Priority Weighting │
└──────────────────────┘     └──────────────────────┘     └──────────┬───────────┘
                                                                     │
                                                                     ▼
┌──────────────────────┐     ┌──────────────────────┐     ┌──────────────────────┐
│ Export Reports       │◀────│   Streamlit Web UI   │◀────│ Local SQLite Storage │
│ (CSV, JSON, Markdown)│     │ & Matplotlib OOP Ax  │     │ (~/.stackcheck/*.db) │
└──────────────────────┘     └──────────────────────┘     └──────────────────────┘

Component breakdown

• src/stackcheck/client/hiringcafe.py: Requests payloads with pagination, workplace filters, and retry logic.
• src/stackcheck/client/normalizer.py: Title categorization, geographic aliases, and duplicate fingerprinting.
• src/stackcheck/analyzer/rule_extractor.py: Positional regex extractor separating requirements from general text.
• src/stackcheck/analyzer/metrics.py: Skill co-occurrence matrix, salary distributions, and weighted scoring.
• src/stackcheck/storage/repository.py: Multi-project SQLite repository managing schema migrations and upserts.
• src/stackcheck/launcher.py: Desktop launcher with free port selection and browser launch logic.
• src/stackcheck/updater.py: GitHub Releases update checker with timeout protection.

Methodology

1. Dataset cleaning: Drops duplicates and spam by generating a title-company-location composite hash.
2. Category classification: Groups technologies into categories (Languages, BI Tools, Databases, Cloud and DevOps, AI and Machine Learning).
3. Requirement weighting:
   • First requirement item: 1.8x
   • Second requirement item: 1.5x
   • Third requirement item: 1.3x
   • Day-to-day responsibilities: 1.2x
   • General post text: 1.0x
   • Preferred qualifications: 0.7x
4. Regional comparisons: Measures how tool adoption varies across countries and regional markets.

Installation options

Option A: 1-line script (Recommended)

# Installs to ~/.local/bin/stackcheck with desktop shortcut and icon
curl -fsSL https://raw.githubusercontent.com/haydermuhib/StackCheck/main/install.sh | bash

# Launch dashboard
stackcheck

Option B: Python source setup

git clone https://github.com/haydermuhib/StackCheck.git
cd StackCheck

python3 -m venv .venv
source .venv/bin/activate

pip install -e .

streamlit run app.py

Option C: Build standalone executable

# Build standalone folder distribution
python build_app.py --onedir

# Or build single-file binary
python build_app.py --onefile

Command line interface

# Launch web dashboard
stackcheck

# Print version
stackcheck -v

# Check status of running dashboard process
stackcheck status

# Verify runtime dependencies
stackcheck check

# Stop running dashboard process
stackcheck stop

# Check for updates and upgrade in-place
stackcheck update

# Query jobs for a role and location
stackcheck search "Data Analyst" --location "United States" --limit 30

# Query remote backend engineering jobs
stackcheck search "Backend Engineer" --workplace remote --limit 50

# Export dataset
stackcheck export --format all

Running tests

Run the unit test suite:

pytest tests/

Quick reference

License

This project is licensed under the MIT License.