--- name: tailor-resume description: Produces a one-page Markdown or LaTeX resume tailored to a specific internship, co-op, or new-grad SWE / data science / ML posting using only facts from the student's profile, with every bullet verified against facts.json. Use when the user asks to tailor, rewrite, or generate a resume or CV for a job. --- # tailor-resume ## Inputs - Target role from `profile/roles.md` (JD + `### Fit` block), or a pasted JD. - `profile/cv.md` and `profile/facts.json` (source of truth). ## Steps 1. Extract the JD's top 6–8 keywords (languages, frameworks, ML/data tools, responsibilities). 2. Select content: education first for students; then the 2–3 most relevant experiences/projects by keyword overlap. Drop weak items to stay on one page. 3. Rewrite bullets: action verb + what + tool + result. Reuse metrics **only** if present in facts.json, with their exact names and units (a crash-free rate stays a crash-free rate). Mirror JD wording only where the fact supports it (e.g. "REST API" if FastAPI service exists). 4. Reorder the skills line to put JD-matching, evidenced skills first. Never add a skill not in `facts.json.skills` or a project stack. A skill that appears only in `skills` stays on the skills line; bullets and summaries claim hands-on use only for skills that appear in an experience or project. 5. **Verify:** produce a mapping table — each bullet → source fact id + the exact supporting string (`exp1: "pytest coverage 55% -> 80%"`). Any bullet without a source is removed or flagged as `[NEEDS FACT]` and you ask the user. 6. Output to `output/resume-.md` (default) or `.tex` (simple one-column LaTeX, no exotic packages). Keep to one page: ≤ ~550 words, ≤ 4 bullets per entry. 7. Show a short diff summary: what moved up, what was cut, keywords covered / still missing (missing ones go back to `score-fit` gaps, not into the resume). Tell `track-pipeline` the resume path for this role_id. ## Rules - **No invented experience, employers, dates, metrics, or skills.** Rephrasing is fine; fabrication is not. - Keep contact info and education dates exactly as in facts.json. - ATS-friendly: no tables/columns/images in the resume itself; standard headings.