jobwatch

jobwatch is an open-source tool for monitoring new job postings on company career pages. It reads public job information from the company job boards you follow (such as Greenhouse, Lever, Ashby, Workable, Workday, Eightfold, Rippling, etc.) and generates a short summary of new jobs ranked by match. It is only responsible for discovery and ranking, and will never automatically apply for jobs for you.

Purpose

jobwatch is designed to help job seekers efficiently track the latest job postings at companies they are interested in. It monitors company job boards directly, so job information is complete and fresh—you can see jobs as soon as they are posted, without having to wait for aggregator sites to index them. It can also read salary ranges from the structured fields of job boards and supports filtering by location, department, salary, posting time, and other conditions.

Main capabilities

• Monitor multiple recruiting platforms: Supports Greenhouse, Lever, Ashby, Workable, Workday, Eightfold, Oracle Recruiting Cloud, SmartRecruiters, Jibe, Rippling, Avature, etc., as well as the recruiting pages of Google and Amazon and Hacker News' "Who is hiring?" posts.
• Job ranking and matching: Ranks jobs according to the filters you set (job title, location, department, salary, posting time, keywords, etc.), and can optionally use a resume for match scoring (requires setting resume and scoring.top or --score N).
• Browser interface: Provides a local web page where you can add companies to monitor, set filters, and view daily selected jobs (up to 10, sorted by best match, newest, or highest salary). Supports queue management, application tracking, a chat assistant, and other features.
• Command-line interface: Provides commands such as find, run, fetch, show, and score, making automation easy (for example, running daily via cron).
• Application tracking: Records each job you have applied to and its status (applied, screening, interview, offer, rejected, withdrawn), and sets next steps and follow-up dates.
• Chat assistant: Provides chat functionality on the Today page or next to each job's details, allowing you to ask questions about today's jobs or a specific job.
• MCP integration: Can be used from Claude via MCP.
• Other features: Supports extracting jobs from Hacker News' "Who is hiring?" posts; supports salary estimation (based on Levels.fyi public salary pages, disabled by default); supports handling situations where Ashby job boards are offline.

How to use

1. Install and run: Get the code from the GitHub repository. Open the local page in a browser, add companies (enter a name or paste a job link), set filters, and then click "Check for new jobs". The page will automatically check every 6 hours (configurable via fetch_every).
2. Command line: Use jobwatch find to find company job boards, add the results to the watchlist file under companies: , adjust the filters, and then run jobwatch run -o ~/jobs-today.md(Can be added to cron to run daily).
3. Configuration: All settings (monitored companies, filter criteria, keywords, resume, chat model, theme, etc.) are in the watchlist file and can be edited via the Settings page. State data (jobs seen, applied to, skipped, and their scores) is stored in a SQLite file, with the default path being ~/.local/share/jobwatch/state.db.
4. Resume scoring: Set resume and scoring.top (or --score N), and jobwatch will use shortlist-ai to match job requirements against the resume and store the scores. The local backend takes one to two minutes per job, so it is recommended to set the top value relatively small.
5. Persistent browser: On Mac, it can be set as a login item so that the page is always available and Today is automatically kept up to date.

Use cases

• Job seekers who want to discover new jobs at target companies as soon as possible without relying on third-party aggregator sites.
• Need to filter and rank a large number of jobs based on resume match.
• Want to track application progress and manage interview preparation.
• Users who prefer to operate via the command line or a chat interface.
• Integrate job monitoring functionality in Claude via MCP.

Limitations

• Does not automatically apply for jobs; it only provides discovery and ranking.
• Does not scrape LinkedIn or Indeed (these sites prohibit scraping).
• When using SmartRecruiters, it does not directly call its API host (prohibited by robots.txt), but instead reads public job board pages.
• For Avature portals, it only reads public pages allowed by robots.txt.
• Regular expressions in title filtering cannot be used for searching; ordinary keywords must be retained.
• Resume scoring is slow when using the local backend. When using the Claude backend, each scoring is a paid API call.
• Users must ensure on their own that their use of data sources such as Levels.fyi complies with their terms.
• On some job boards (such as Ashby), job pages may display "Page not found"; in this case, the company's job site must be manually added to the configuration.
• Hacker News posts are free text and may not be accurately parsed into specific jobs, and resume scoring is not performed.
• When setting ANTHROPIC_API_KEY, if the page is started via launchd, additional environment variable configuration is required.