Score it, see what the job description wants that it does not say, fix the wording with the verb and skill lookups, then export the ATS PDF. Because it writes to the same store as the web editor, nothing has to be re-entered on the other side.
The stdio MCP server for JuicedResume. It reads and writes the same store as the web editor, so a change made from the client shows up there immediately and vice versa. The surface covers the whole document — resumes, personal info, sections, entries, template and styling — plus scoring, JD matching, lookups, and export to LaTeX or an ATS-friendly PDF.
- Resumes listed, read, created blank, duplicated and deleted
- Personal info patched key by key, and the summary section's body replaced
- Sections added by type — summary, experience, education, skills, languages, certificates and more — then removed, renamed, reordered and hidden
- Entries added, patched and removed inside a section
- Template switched, and styling patched: accent, fonts, spacing, layout, photo
- Locale patched — language, date format, page format
- Resume Worded-style scoring
- A resume matched against a job description, returning a relevancy score and the missing keywords
- Fuzzy lookups over action verbs by category, the skills dictionary, job titles and company names
- Export to LaTeX source, or compiled to an ATS-friendly PDF saved to a path or returned as base64
A JuicedResume account and an MCP token from Settings > AI, placed in the server's environment; MCP access is included on the Basic, Pro and Lifetime plans. Node.js, then npx -y @juicedresume/mcp or a global npm install. Install the published package rather than building this repository — the scoring engine and resume schema live in the JuicedResume monorepo.
One command — npx -y @juicedresume/mcp