jd-resume-match

jd-resume-match is a Chinese-language job-search Skill by 欧八同学. You give an AI agent your resume and one or more job descriptions. The AI maps each requirement to evidence in your resume, asks you about anything unclear and suggests structural changes, while scores, unknowns, apply-or-not advice and rankings come from Python scripts with fixed rules. The output is a self-contained HTML report, plus an optional PNG share card and post text; nothing is posted automatically. The README says the score measures how much evidence your materials provide, not your ability or your chance of passing an interview. When it revises a resume, it only rearranges facts you already have, every number has to be traceable, it adds no new responsibility words such as "independently" or "led", and missing information is marked unknown.

Who it's for:People weighing several openings at once who want to see how well their resume matches each one. You need a tool such as Claude Code or Codex CLI to run it.

Before you start: Install with git clone or by copying the folder. The optional install.sh copies the folder and runs local validation scripts automatically; by default it will not overwrite an existing non-empty folder, and --force deletes and replaces that install folder. You need Python 3.9 or later, standard library only, with no pip or npm packages. PNG export uses a Chrome, Chromium or Edge browser already on your machine, with a separate temporary profile that leaves your everyday logged-in browser alone. Chinese PDFs with CID fonts and scanned files are not supported; use DOCX or TXT. It works with Claude Code, ZCode, Codex CLI and similar tools, mostly in Chinese. The resume analysis goes through whichever AI model you use. SKILL.md says the offline HTML only means you can view the report without internet, not that the analysis runs offline, and the Skill never asks for or stores API keys. If you paste a job link, it tries a public fetch; pages behind a login or anti-scraping protection fail, and you paste the full text instead. The full report contains your private details. The share card goes through a redaction scan, and you should preview it before posting. The Skill is free; model costs depend on your tool. MIT license.

Based on the project's public documentation. Not tested by AI Inspiration World.