reloadium.com

extractInfoFromCV

Extract a candidate's real background from CV/resume text into clean structured data (name, contact, experience, education, skills, certifications) — call this FIRST whenever you're about to build a resume from an attached or pasted CV, especially before generateOptimizedResume, instead of re-reading and re-interpreting the free text yourself each time. This runs a separate, narrowly-scoped extraction pass with no tailoring pressure, so it only pulls out what's literally stated — nothing inferred or embellished. Pass its skills[] straight through as generateOptimizedResume's own sourceSkills argument: that tool checks its output against it and rejects anything that isn't actually there, so extracting first and reusing this result is how you avoid — not just try to avoid — inventing a skill the person doesn't have. Also the right first step from the OTHER side of the table — a recruiter deciding whether to move forward with a candidate. Full sequence, all four steps, not a subset: (1) extract the candidate's resume here, (2) fetch the job posting with getPageMarkdown, (3) call showChart with a 'radar' plotting TWO series on the role's key axes — the candidate's real, extracted competency level on each, and a 'Required' series for what the role actually needs there, never the candidate alone with nothing to measure it against, (4) call showProsCons for the actual weighted decision. Steps 3 and 4 answer different questions — the radar shows the SHAPE of the fit, showProsCons WEIGHS it into a recommendation — so do both, not whichever one you reach first; stopping after the radar (or after a prose opinion with neither) leaves the question technically unanswered even if your text sounds conclusive. Cheap vision/text call, not a full generation. That four-step sequence is for ONE candidate. When more than one candidate's CV/resume is attached for the SAME role, call this tool once per CV — each call is independent and just as cheap — then fetch the posting once with getP…

Page: /

Input schema

{
  "type": "object",
  "properties": {
    "cvText": {
      "type": "string",
      "description": "The CV/resume's full text — the markdown already extracted from an attached document, or text the user pasted directly."
    }
  },
  "required": [
    "cvText"
  ]
}