Help center
Start with synthetic content, understand the two analysis modes, and use every opportunity as a prompt for truthful human review.
Guided answers
18 answers
Add résumé content, add the complete job description, choose an analysis mode, and run the scan. Review the score in the context of its denominator, inspect curated relevant terms, classify truthful opportunities, then export or edit and rerun.
Load the synthetic demo →Skills-focused is the default curated relevance view. It uses only documented stop words, phrases, synonyms, and categories. Full lexical preserves the v1.0 unique-token score as a raw compatibility baseline and separately surfaces curated Relevant Keyword Review terms.
Read the methodology →The denominator depends on the selected mode. In Full lexical it means two of three equally weighted unique job-description tokens matched. In Skills-focused it means two of three recognized categorized concepts matched. It does not mean a 66.7% chance of passing an ATS or being hired.
Skills-focused opportunities are supported categorized concepts present in the job description but absent from the combined résumé input. Full lexical instead shows factual exact shared and unmatched raw tokens. Relevant Keyword Review separately filters to curated categorized concepts. Add language only where it truthfully reflects your experience.
Relevant Keyword Review is a deterministic curated layer shown separately from the raw Full lexical score. It keeps only explicitly supported categorized concepts and aliases, preserves exact source evidence, and marks them represented or missing. It does not infer experience, requirement severity, or application readiness.
See the scoring boundaries →The curated taxonomy groups explicitly supported concepts. Uncategorized keeps unknown meaningful terms visible but excludes them from the primary Skills-focused score. The interface shows the recognized curated-concept denominator so category coverage is not presented as whole-job-description coverage.
Use searchable PDF, DOCX, or UTF-8 TXT files up to 10 MB each. The request can include up to five résumés, 25 MB of files in total, and 200,000 characters per pasted field or extracted document. Image-only PDFs need OCR before upload.
Résumé files and pasted résumé text are combined into one comparison corpus. Pasted job-description text takes precedence over an uploaded job description. The interface labels both rules before submission.
The API safely rejects unsupported, empty, oversized, malformed, encrypted, image-only, binary-text, or suspicious DOCX archive content. Convert scans with OCR, remove a password, or save a clean supported file before retrying.
The public frontend sends them to the configured Render API for the current request. The application does not intentionally persist document content, closes upload resources, marks analysis responses no-store, and excludes all document and form content from analytics.
Read the privacy details →Classify each curated review term as Add to résumé, Already represented, Not relevant, or Review later. Reviewed means you made a decision—not that the résumé is optimized. Full lexical raw unmatched tokens remain visible separately and are not automatically treated as résumé opportunities.
Edit either input and the existing result becomes visibly stale; export and review decisions are disabled until you run a fresh scan. New analysis requires confirmation and clears the current in-page session.
Download a Unicode-safe server-generated PDF for the current input signature, print the visible results, copy a factual summary or selected terms, or download a plain Markdown checklist. Temporary browser download URLs are revoked after use.
The free API service can cold-start after inactivity. The client uses a bounded 45-second request, preserves inputs, supports cancellation, and offers manual retry. It never invents progress stages or automatically repeats a POST request.
Your successful analysis remains visible. Retry the report from the export section. Changing inputs or starting a new analysis cancels an in-flight report so an old document cannot download after the session is cleared.
Yes. Press Command-K or Control-K for available page actions. In the review queue, focus a row and use Up/Down to move, Enter to inspect its status, or 1–4 to apply the four labeled decisions. Standard Tab, Shift+Tab, Enter, Space, and native form-control behavior remain available. Escape closes menus, the command palette, and confirmation prompts. Results receive focus after a successful scan, and reduced-motion and forced-color preferences are respected.
Lexical means text-based. Deterministic means the same inputs and rules produce the same result. Normalization means only explicitly documented forms—such as QC and quality control—map to one concept. No AI, fuzzy matching, or hidden hiring model is used.
Use the public GitHub issue form with synthetic examples only. GitHub stores submitted issues under its own account and retention policies; the screener itself stores no feedback. Never include résumé text, job-description text, filenames, credentials, or other personal information.
Open a GitHub issue ↗