Ron-Salama

Jobscan — DevOps skill for Claude Code

DevOps community

Job-scanning data pipeline: ~30 Israeli job sources (REST/JSON APIs, ATS platforms, scrapers), a tested bilingual rule engine, dedup and a tracker page.

How to install Jobscan

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Ron-Salama/jobscan and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Jobscan does

Job-scanning data pipeline: ~30 Israeli job sources (REST/JSON APIs, ATS platforms, scrapers), a tested bilingual rule engine, dedup and a tracker page. GitHub Actions CI/CD, 64 regression tests, Claude Code judge + skeptic AI agents. Python.

Alternatives in DevOps

  • Open Agent Email — Open-source email infrastructure for AI agents: secure inboxes, REST and MCP APIs, SMTP, human approval workfl 2.1k ★
  • Skill — Web extraction engine with antibot bypass 549 ★
  • Claude Codex Settings — by fatih akyon - A well-organized, well-written set of plugins covering core developer activities, such as wor 510 ★

README

JobScan

A job scanner I built for my own job search. It reads about 30 Israeli job sources, filters them with tested rules in English and Hebrew, merges duplicates across sources, and publishes one tracker page. Claude Code agents wrote most of the code. I set the requirements and acceptance criteria, made the rule decisions, and checked the agents' work against reality.

[![tests](https://github.com/Ron-Salama/jobscan/actions/workflows/tests.yml/badge.svg)](https://github.com/Ron-Salama/jobscan/actions/workflows/tests.yml)

**Stack:** Python · data pipeline over REST/JSON APIs, ATS platforms and HTML scrapers · bilingual (English + Hebrew) rule engine · GitHub Actions CI/CD with regression tests gating every run · monitoring and alerting (source-health alarm, Telegram) · Claude Code multi-agent review (judge + skeptic sub-agents) · CLAUDE.md agent context

**Live tracker:** [ron-salama.github.io/jobscan](https://ron-salama.github.io/jobscan/) (refreshed automatically after every cloud scan)

![JobScan tracker: the public view of the production page](docs/screenshot.png)

The public view of the production tracker (`docs/index.html`, 499 roles on 2026-10-04), rebuilt and pushed here by the private deployment after every run: the real listings with the verdicts and scores from the AI review step. My application statuses, notes, pay estimates, CV labels and verdict reasons are removed. Built by `tools/build_snapshot.py`.

What it does

One production run (2026-10-04, 17 minutes on GitHub Actions):

Step Roles
Raw postings from 30 sources 7,256
Pass the rules (region, title gate, seniority, years) 483
After cross-source dedup 478
New since the last run 10
On the tracker page (it also carries roles from earlier runs) 500

Roles the rules are unsure about go to three buckets instead of being dropped. In that run the giant bucket kept 1,587 of 1,588 (one duplicate URL) and the QA bucket kept 170 of 186 (16 were