ResumeSkills: Claude Code Skills for Resume Optimization and Job Applications

What ResumeSkills actually is, and who it is for

ResumeSkills is not an application. It is a folder of markdown files that an AI coding agent reads as instructions. The README defines skills as "markdown files that give AI agents specialized knowledge and workflows for specific tasks", and the repository ships 20 of them under skills/, each with its own directory such as resume-ats-optimizer, resume-bullet-writer and job-description-analyzer.

The target user is narrow and specific: someone who already has Claude Code, Cursor, Windsurf, Codex or Gemini CLI open in a terminal or editor and wants the assistant to stop giving generic resume advice. Once the skills are in place, the README's usage examples show plain requests mapping to specific skills. Typing "Optimize my resume for ATS" is documented to trigger resume-ats-optimizer; pasting a job description alongside "Should I apply to this job?" is documented to trigger job-description-analyzer.

The repository also carries per-agent configuration directories at the top level: .claude/, .cursor/, .codex/, .gemini/, .windsurf/, .opencode/, .agent/ and .agents/. That layout tells you the author intends the same skill text to be consumed by several different agent runtimes rather than one vendor. For a job seeker, that matters less than it sounds. What matters is that the skills are plain text you can read and edit before the agent ever applies them.

How the skill mechanism works inside your agent

There is no server, no API and no build step. The mechanism is file discovery plus prompt injection. Each skill lives in its own directory under skills/, and the agent decides from your request which directory is relevant, loads its contents into context, and follows the workflow described there.

The README describes this as recognition rather than routing: "Claude Code can recognize when you're working on resume and job search tasks and apply the right frameworks and best practices." That is a meaningful distinction. Nothing in the repository enforces which skill fires. The selection is the model's judgement, which means two runs of the same request can pull in different skills, or none.

Because the payload is prose, the quality of the output tracks the quality of the skill file. A skill that says "add metrics" produces different results from one that specifies which bullet patterns to rewrite and which numbers to ask the user for. The README lists resume-quantifier as a skill that finds opportunities to add metrics and estimates them "when numbers unknown", which is exactly the kind of instruction you should read before trusting the output. An agent estimating a number for your resume is a decision you want to make consciously.

The skills are grouped in the README into five categories: resume optimization, job search strategy, supporting documents, interview and negotiation, and specialized roles. The grouping is documentation, not architecture. All 20 are peers on disk.

Installing ResumeSkills with npx and running a first ATS check

The README gives three install paths. The CLI route is the recommended one and installs all 20 skills at once. The -g flag targets your global skills directory so the skills are available across every project; dropping -g installs into the current project only.

npx skills add Paramchoudhary/ResumeSkills -g -y

After it finishes, the README documents a listing command to confirm what landed:

npx skills list --global

If you would rather not run the CLI, the manual route clones the repository and copies the skill directories into an agent's skills folder. The README's example targets Cursor:

mkdir -p ~/.cursor/skills
cp -r ResumeSkills/skills/* ~/.cursor/skills/

For a first real use, open your agent in a project that contains your resume as a text or markdown file and make the request the README documents for ATS work. The expected behaviour is that the agent loads resume-ats-optimizer and walks through compatibility and keyword matching against the job description you supply.

Optimize my resume for ATS

What you should see is a structured pass over formatting and keyword coverage, not a rewritten resume in one shot. Read the skill file at skills/resume-ats-optimizer first so you know what the agent was told to check. Uninstalling is per-skill by name, or by deleting the copied directories:

npx skills remove resume-ats-optimizer

The ATS claim is the weakest part of the pitch

The README's rationale section leads with "75% of resumes rejected by ATS before humans see them" and closes with results including "2-3x more interviews per application" and "$10K+ higher offers". None of those figures carry a source, a sample, or a methodology in the README, and no release notes or benchmark data are present in the repository to back them. Treat them as marketing copy, not measurement.

The technical problem underneath is real. Applicant tracking systems are not one system. They parse, rank and filter differently, and a markdown skill file cannot know which vendor will read your document. What resume-ats-optimizer can plausibly do is check the things that are universally safe: standard section headings, keyword presence against a specific posting, and formatting that survives plain-text extraction. What it cannot do is tell you your score in a particular employer's system, because the skill has no access to that system.

There is a second failure mode worth naming. The skills work on text you paste into the agent's context. If you paste a resume containing a home address, a phone number or a current employer's internal details, that content goes wherever your agent's requests go. The README says nothing about data handling, and no privacy or redaction guidance appears in the documented workflow. If your resume contains anything you would not paste into a third-party tool, redact it before the first request.

Where the 20 skills overlap with each other

Twenty skills sounds like breadth until you read the list closely. resume-ats-optimizer, resume-formatter and resume-section-builder all touch layout and structure. resume-bullet-writer and resume-quantifier both rewrite the same bullet lines, one for achievement framing and one for numbers. resume-tailor and job-description-analyzer both consume a job posting, with tailoring downstream of analysis.

That overlap is not automatically a defect. Splitting bullet rewriting from metric estimation lets you run one without the other, which is reasonable if you want to add numbers without changing phrasing. But it does mean the agent's skill selection has more room to pick the wrong one, and the README's own examples show single requests mapping to single skills with no documented way to force a specific skill by name. If your agent picks resume-formatter when you wanted resume-ats-optimizer, the README does not document an override.

The specialized skills are the more interesting half. tech-resume-optimizer, executive-resume-writer, academic-cv-builder and creative-portfolio-resume each encode conventions for a distinct audience, and creative-portfolio-resume explicitly has to balance visual design against ATS compatibility, which is a genuine tension rather than a formatting preference. If you are in one of those four buckets, the specialized skill is more likely to be worth your time than the generic optimizer.

ResumeSkills versus a hosted resume builder

The obvious alternative is a hosted resume service such as a paid ATS scanner or an online resume builder. The difference in approach is architectural, not cosmetic. A hosted scanner owns the parsing engine and can return a deterministic score against its own rules. ResumeSkills owns nothing; it hands instructions to whatever model your agent is running, and the output varies with that model, your prompt, and the context window.

That trade runs both ways. A hosted tool gives you a repeatable number and no ability to change how the number is computed. ResumeSkills gives you a markdown file you can open, disagree with, and edit, and no number at all. If you want to argue with the tool's definition of a good bullet, the skill file is the argument.

The other real difference is cost and data path. The skills are MIT licensed and run inside an agent you are already paying for, so there is no separate subscription. The counterweight is that your resume text enters your agent's request pipeline rather than a vendor's resume-specific system. Neither path is private by default. The question is which one you have actually read the terms for.

A third option, and a legitimate one, is hiring a human resume writer. That is slower and more expensive, and it produces a document rather than a reusable workflow. ResumeSkills is the better fit if you are applying to many roles and want the per-application tailoring to be cheap.

Licence, maintenance and what upgrades cost you

The repository is MIT licensed, which permits commercial use, modification and redistribution provided the copyright notice and permission notice are retained. That is permissive enough that the licence is unlikely to be the reason you say no. It is not legal advice; if you plan to redistribute the skills inside a product, read the LICENSE file in the repository root.

The last push to the default branch was on 2026-06-19. No releases are present in the repository, so there is no versioned upgrade path and no changelog to read before pulling changes. Upgrading means re-running the install command or re-copying the skills directory, and the practical cost is that any local edits you made to a skill file will be overwritten unless you keep your own copy. If you customize skills/resume-bullet-writer to match your industry's phrasing, that edit lives outside the project's version control.

The maintenance cost on your side is close to zero as long as you treat the skills as text. There is no dependency to update, no runtime to patch, and no service that can go down. The failure mode is quieter: an agent that silently stops matching your request to a skill produces generic advice, and nothing in the repository alerts you when that happens. The README does not document any diagnostic for skill selection.

Editorial conclusion

Adopt ResumeSkills if you already work inside Claude Code, Cursor or a similar agent and want structured prompts for ATS checks, bullet rewriting and job-description analysis without paying for a resume service. Skip it if you need deterministic ATS scoring, a hosted editor, or any guarantee about the outcome numbers the README quotes. Before relying on it, open skills/resume-ats-optimizer and skills/job-description-analyzer and read the actual instructions the agent will follow, because that text is the product.

Frequently asked questions

Official sources

1. Issues
2. License: MIT
3. Paramchoudhary/ResumeSkills on GitHub
4. README