offer ready GitHub

01 One pipeline · two open components

Turn a job link into a CV worth sending.

Extract the real requirements. Match only what your experience can prove. Compile a tailored LaTeX CV—and verify the result before it leaves your machine.

Start with one command Read the source
  • Read-only MCP
  • No invented experience
  • MIT licensed

Understands job pages from

LinkedIn+InfoJobs+Indeed+ TecnoEmpleo+InfoEmpleo+the open web

02 Use either. Combine both.

Two focused tools.
One honest application.

Each component solves one job well. Together, they close the gap between a job board and a finished, machine-readable CV.

01 MCP SERVER

THE EXTRACTOR

Give your agent the job offer—not the page clutter.

One read-only tool turns public URLs into consistent descriptions and criteria. Dedicated extractors cover major job boards; safe generic retrieval handles the rest.

  • Structured, predictable output
  • Public-URL validation
  • Stable, explicit error codes
mcp.jsonREADY
{
  "command": "uvx",
  "args": [
    "job-offer-scraper-mcp"
  ]
}
02 CODEX SKILL

THE TAILOR

Tailor the message. Keep every claim true.

The skill maps requirements to evidence already present in your LaTeX CV, rewrites only what can be supported, then compiles and inspects the PDF locally.

  • Evidence-based keyword matching
  • Pinned, hash-verified compiler
  • Extractable PDF text checks
REQUIREMENT MATCH3 / 4
PythonDIRECT
REST APIsDIRECT
AWSEQUIVALENT
KubernetesOMITTED

Unsupported requirements stay out of the CV.

03 From URL to verified PDF

A pipeline you can inspect.

  1. 01
    Fetch

    The MCP validates and reads the public job URL.

  2. 02
    Structure

    Role, responsibilities and criteria become agent-ready data.

  3. 03
    Match

    Every requirement is compared against evidence in your CV.

  4. 04
    Tailor

    Relevant language moves forward. Unsupported claims do not.

  5. 05
    Verify

    LaTeX compiles locally; the final PDF is checked before publication.

Optimise the presentation.
Never manufacture the person.

LOCAL FIRSTYour CV stays on your machine throughout the workflow.

VISIBLE BY DESIGNNo hidden keywords, invisible text or ATS tricks.

FAILS EXPLICITLYIf an offer cannot be read or a PDF cannot be verified, you know.

04 Open source · ready to run

Start with the extractor.

Python 3.11+. No clone required. Run the published package in an isolated UV environment.

TERMINAL01 / 02
$ uvx job-offer-scraper-mcp
INSTALL THE SKILL FROM02 / 02
github.com/JuanjoLopez19/job-scrapper-mcp/tree/master/skills/tailor-latex-cv