# Cv Engineering > CV Engineering - Skill: `handsomeboy990/cv-engineering` (Agent Skill, multi-file: 4 files) - Install (CLI): `npx skillmds@latest add handsomeboy990/cv-engineering` - Raw SKILL.md: https://api.skillmd.com/api/skills/handsomeboy990/cv-engineering/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: Handsomeboy990 (https://skillmd.com/u/handsomeboy990) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/handsomeboy990/cv-engineering --- # CV Engineering A CV has two readers: a machine that parses it in a second, and a human who scans it in six. It has to survive both, and it has to be true. This skill renders the candidate's real profile into a document that passes the parser, holds the human, and states nothing the candidate cannot support. ## 1. The format the market expects ``` target the format the candidate's market and level expect: the conventions differ by country, industry, and seniority, and are not assumed ats a layout an applicant-tracking system can parse: real text not images, standard section headings, no tables or columns that scramble on extraction, a common font, standard bullet characters scan a human's six-second scan lands on the name, the current role, and the top achievements; the layout puts those where the eye goes length one page early-career, two mid-to-senior, rarely more; length is earned by relevance, never by padding ``` ## 2. Achievement bullets that quantify The unit of a strong CV is the achievement bullet, and the weak version is a duty. ``` duty "Responsible for the billing system" (what the job was) achievement "Rebuilt the billing pipeline, cutting failed charges 40% and month-end close from 3 days to 1" (what changed) formula action verb + what was done + the quantified result, true number a real number from the profile; if none exists, a real qualitative outcome, never an invented figure relevance the bullets chosen are the ones that matter to the target role, not the candidate's favourites ``` ## 3. Tailoring without invention ``` emphasise for a specific role, lead with the real experience that matches it; the same true history, ordered to the role language mirror the role's real vocabulary where the candidate's experience genuinely fits it, so the parser and the human both recognise the match never add a skill, tool, or result the candidate does not have to match the listing; tailoring reorders and reframes truth, it does not add to it cut the experience irrelevant to this role, to make room for what matters ``` ## 4. Relevance and length, ruthlessly ``` keep what advances the candidate's case for this role cut old roles at bullet length once they are far behind; a line each is enough compress early-career detail as the career grows; the graduate project leaves when the senior achievements arrive zero no objective-statement cliches, no skill bars, no "references available", no filler that survives only because it is traditional ``` ## 5. The parsing pass ``` extract run the CV as a machine would: does the text extract cleanly, do the sections parse, do the dates and titles come through keywords the real, relevant terms from the target role appear where the candidate's experience genuinely supports them, so a keyword filter does not drop a qualified candidate contact name and contact details parse correctly; a CV that scores well and loses the phone number has failed ``` ## 6. Prohibitions - Never state a skill, title, result, date, or tool the candidate cannot support. - Never invent a number; use a real one or a true qualitative outcome. - Never add to the truth to match a listing; tailoring reorders, it does not fabricate. - Never use a layout that a tracking system cannot parse (image text, complex columns) if the candidate's market uses such systems. - Never pad length; relevance earns length. - Never keep a cliche (objective statement, skill bars, "references available") because it is traditional. ## 7. Protocol 1. Read the real profile from `career-profile` and the target role. 2. Choose the market- and level-appropriate, parseable format. 3. Write achievement bullets: action verb, what was done, quantified true result. 4. Order and emphasise to the target role, adding nothing untrue. 5. Cut to relevance and to the right length. 6. Run the parsing pass: clean extraction, real keywords, correct contact details. 7. Verify every line is supportable in an interview, per career-core. ## 8. Auto-critique Score from 0 to 5: format appropriate and parseable, bullets are quantified achievements not duties, tailoring reorders truth without adding to it, ruthless relevance and correct length, parsing pass clean with real keywords, every line supportable, no cliche filler. Threshold: no axis below 3, average at least 4. A single line the candidate cannot support, or an invented number, caps the score until corrected, per career-core. ## 9. Interfaces - Upstream: `career-core` for honesty, `career-profile` for the real material, `job-search` for the specific role to tailor to. - Downstream: `cover-letter` complements the CV without repeating it, `interview-preparation` drills the achievements the CV claims. - Lateral: `document-design` and `pdf-production` in `documents/` render the CV to a polished, correctly paginated PDF; `administrative-writing` for a formal application package. ## 10. Note on rendering This skill produces the CV's content and structure. When a polished PDF is required, the rendering (typography, pagination, a clean single- or two-column layout that still parses) is handed to `document-design` and `pdf-production`, which verify the rendered pages rather than trusting the generator. A CV that looks perfect and does not parse is a failure this handoff exists to prevent.