jd-match
Match a resume against a specific job description, giving a transparent match score, requirement-by-requirement evidence map, gaps sorted into "fix on resume" vs "learn in 7 days" vs "can't close quickly", tailored bullets (no fabrication), and a 7-day plan to close the gaps before applying or interviewing. Use when the user pastes a job description or JD link, asks "am I a fit for this role", wants to tailor their resume to a job, or prepare for a specific company's role.
How do I install this agent skill?
npx skills add https://github.com/nishant-tiwari24/cracked --skill jd-matchIs this agent skill safe to install?
- Gen Agent Trust Hubpass
The skill is functional and poses no direct malicious threat; however, it is susceptible to indirect prompt injection as it processes external resumes and job descriptions without sanitization.
- Socketpass
No alerts
- Snykwarn
Risk: MEDIUM · 1 issue
What does this agent skill do?
JD match
A job description is the exam paper, handed over in advance. This skill reads it properly, maps the candidate's real evidence against it, and turns the gaps into a plan.
0. Inputs
- Call the Skill tool with "cracked-profile" (resume path, level).
- Get the job description: pasted text, a file, or a URL (fetch it if you have a web tool; otherwise ask them to paste it).
- Call the Skill tool with "company-intel" for the JD's company, unless it's unnamed or a staffing agency. Its process and DSA focus feed the "likely interview questions" and 7-day plan below.
- Get the resume (same reading rules as the
roast-my-resumeskill: PDF/DOCX/LaTeX/MD/text). Also accept extra evidence the resume doesn't show ("I also know Docker from my internship"), and note that it isn't on the resume yet.
1. Decode the JD
Extract into a table:
| # | Requirement | Type | Weight |
|---|
- Type:
must(stated as required, or repeated),nice(preferred/bonus), orimplicit(not stated but obvious from the role, e.g. Git, DSA for SDE roles, SQL for backend). - Weight: 3 for must, 2 for implicit, 1 for nice.
- Also pull out the seniority signals (years, "own", "lead"), interview hints (the stack implies likely questions), and ATS keywords, using the exact phrasing in the JD.
Note anything unusual: unrealistic requirements for the level, bond or service-agreement mentions, or a mismatch between title and responsibilities.
2. Evidence map
For each requirement, find the strongest evidence in the resume or extra evidence:
| Requirement | Evidence | Strength |
|---|---|---|
| REST APIs (Node.js) | "Built order APIs…" (Projects › ShopKart) | ✅ strong |
| Docker | mentioned verbally, not on resume | 🟡 hidden |
| Kubernetes | none | ❌ missing |
Strength: strong (demonstrated with context) = 1.0, weak (listed in skills only) = 0.5, hidden (true but not on the resume) = 0.5, missing = 0.
Match score = Σ(weight × strength) / Σ(weight), shown as a percentage with the formula visible, so the candidate can trust it. Then interpret it: ≥ 75% apply confidently · 55–75% apply after the fixes below · < 55% a stretch, so apply only if the "can't close" list is short.
3. Gap triage
Sort every non-strong item into exactly one bucket:
- Fix on resume (today): they have it but it's hidden or weak. Give the exact rewritten bullet or skills line. Only use real experience. Missing numbers become placeholders.
- Learn in 7 days: closable fast enough to talk about credibly (a tool, a library, a concept). For each, name a concrete mini-project or exercise that produces evidence, not just "watch a tutorial".
- Can't close quickly: years of experience, deep specialisations. Advise how to address it honestly if it comes up in the interview.
4. Output
- Match score and interpretation.
- Evidence map table.
- Tailored resume edits (before → after), plus the ATS keywords to add, where true.
- 7-day plan: a day-by-day table mixing gap-closing work with role-specific interview prep. Example: day 3, "Dockerise your ShopKart backend and push the image; add one bullet." Day 6, "Run /mock-interview on graphs, which this team's domain implies."
- Likely interview questions for this JD: 5 technical (from the stack, plus the company's most reported DSA topics) and 2 behavioral (from its values), each labelled with where it came from.
- Save everything to
$CRACKED_DIR/jd/<company>-<role>.mdand mention the path. - Log a scorecard via the Skill tool with "cracked-profile", titled
<Role> @ <Company>. Dimensions: Must-haves, Implicit Skills, Nice-to-haves, Resume Alignment. Overall = match score / 10.
Rules
- No fabrication, ever. Recommending someone claim a skill they don't have gets them rejected in the interview, or fired later.
- Quote the JD's exact wording for keywords. ATS matching is often literal.
Skill calls: if your agent has no Skill tool, open the named skill's SKILL.md in the cracked skills folder (core/, interview/, profile/, strategy/) and follow it instead.
How can the creator link this skill?
Add the canonical catalog link to the repository README so users can inspect current installs and available audits. The publishing guide covers the complete discovery path.
<a href="https://skillzs.dev/skills/nishant-tiwari24/cracked/jd-match">View jd-match on skillZs</a>