Meet Sweepy

Sweepy automates your job search across many hiring platforms. One command gathers open roles and reads your resume. Sweepy then scores each posting against your skills and writes a daily match report.

Sweepy adapts to your path through a config file. Pick titles like Paralegal, Analyst, or Coordinator.

Meet Alex

Throughout this guide, you'll follow Alex. Alex just graduated with a degree in Political Science and wants to find Paralegal and Legal Assistant roles in New York City. Alex has never used a terminal, written code, or even seen a YAML file.

This guide walks you and Alex from setup to a fully automated job sweep. Sweepy scores top-matching roles against a real resume and builds a PDF directory you can use every day.

note

You don't need any programming experience to use Sweepy. This guide explains every step from scratch.

Features

Sweepy gives you everything needed to find your next role faster:

• Multi-platform search scans Greenhouse, Ashby, Lever, and Workable, four applicant tracking systems that host public job boards. It also reads the Built In New York City board and, once you switch them on, the LinkedIn, Indeed, Glassdoor, and ZipRecruiter feeds.
• Smart resume scoring reads your resume PDF or Word file and extracts your skills. Sweepy routes each posting to one resume track, then scores it against that track's resume. The engine runs without API keys or paid services.
• Broken link filtering checks every posting before you see it. Sweepy drops a posting when the link returns a gone status, a soft-404 page, or a connection error. Rate limits, bot blocks, and server errors keep the posting in your report, marked unverified.
• Daily PDF reports compile your results into a high-density PDF directory. Each page holds six cards by default, with clickable links, disclosed pay ranges, and color-coded badges.
• Pipeline tracker spreadsheet exports a CSV with every card field plus a 1,000-character posting excerpt, alongside a JSON file for other tools. The CSV ends with job-tracking columns for date applied, status, interview stage, offer, follow-up date, contact, and notes. Sweepy carries the values you enter into later sweeps of the same profile. Open the CSV in Microsoft Excel, Google Sheets, or Notion. The PDF serves one ranked reading pass with clickable apply links. The spreadsheet records the whole pipeline, including roles past the PDF card limit.

How it works

A successful sweep follows five stages:

1. Load your profile. Sweepy reads your configuration file, which holds your job titles, location, and resume path.
2. Discover roles. Sweepy queries the hiring platforms and job feeds for matching postings.
3. Validate links. Sweepy checks every link and drops postings that return a gone status, a soft-404 page, or a connection error. Postings with unverifiable links stay in the run, marked unverified.
4. Score matches. Sweepy parses your resume PDF or Word file and calculates a weighted match percentage for each role.
5. Generate reports. Sweepy compiles the results into a styled PDF and CSV or JSON exports.

Mermaid (image)

Mermaid (code)

flowchart TD
    A["Load profile<br/>job titles, location, resume path"] --> B["Discover roles<br/>hiring platforms and job feeds"]
    B --> C["Validate links"]
    C -->|gone, soft-404, or error| D["Drop posting"]
    C -->|reachable| E["Keep posting"]
    C -->|rate limit, bot block, 5xx| F["Keep posting, marked unverified"]
    E --> G["Score matches<br/>resume keywords against the posting"]
    F --> G
    G --> H["Generate reports<br/>PDF, CSV, JSON"]

ASCII

ASCII diagram

 [Load profile]
   job titles, location, resume path
        |
        v
 [Discover roles]
   hiring platforms and job feeds
        |
        v
 [Validate links]
        |
        +--> gone, soft-404, or error ---- drop posting
        |
        +--> reachable ------------------- keep posting
        |
        +--> rate limit, bot block, 5xx -- keep it, marked unverified
        |
        v
 [Score matches]
   resume keywords against the posting
        |
        v
 [Generate reports]
   PDF, CSV, JSON

Limitations

Sweepy covers a narrow slice of the job market by design. Pair it with a language model through the Model Context Protocol (MCP) server, and the model takes over the mechanical defaults. When the Solo wizard and MCP columns report the same behavior, the limitation affects both versions of Sweepy.

Limitation	Solo wizard	With the MCP server	Workaround
Match scores	Counts shared keywords between your resume and the posting. The card reads KEYWORD OVERLAP.	submit_judgments sets its own percentage, and the card reads MATCH.	Raise resume.keywords_weight from the Sweep Settings tile, and add resume.synonyms groups by hand. The wizard leaves synonyms to you.
Pay and benefits	Prints a disclosed pay range, read from structured board fields or the posting text. Benefits always read Not disclosed.	submit_judgments supplies either value.	Fill the comp and benefits columns in your CSV export.
Badges and app format	Infers Amber or Slate from a fixed keyword list in the engine.	submit_judgments overrides both fields.	Check the badge against the live posting before you trust it.
Summaries, requirements, and application tips	Extracts the text from the posting and scores it against your resume.	submit_judgments rewrites them, and render_report accepts a custom playbook.	Edit the fields in your CSV or JSON export.
Resume text	Reads the text layer of PDF and Word files. A scan returns nothing to score.	The model reads the scan with vision and passes the skills in.	List your skills in resume.must_have_skills. The wizard's must-have skills step covers this.
Coverage	Queries 41 named company boards plus 5 Built In NYC categories.	import_jobs appends roles from anywhere.	Add targets to sources.greenhouse.boards, sources.ashby.orgs, sources.lever.companies, or sources.workable.slugs.
Job feeds	Off until you enable them. Each feed returns up to 25 postings per title for the last 3 days.	Same switch.	Toggle a site under sources.jobspy, such as glassdoor.
Link checks	A connection error prunes the posting.	Same.	Keep the JSON export and recheck your top targets on the company site.
PDF length	Caps the PDF at 24 cards.	Same.	Set report.max_cards. A value of 0 removes the cap, and your CSV and JSON exports list every role.
Scraper fragility	Breaks when a board changes its markup or tightens its rate limits.	Same.	Run with --debug, then --logs, to list source errors. A source error never stops a sweep.

Guide roadmap

This guide walks through four steps:

1. Install Sweepy on your computer. See Installation.
2. Configure a personalized job search profile, like Alex's New York City paralegal profile. See Setting up Sweepy.
3. Run your first automated job sweep and watch it work. See First Sweep.
4. Read your custom PDF report and understand match scores, badges, and application strategies. See Understanding Results.