Skip to main content
Glama

jobkit-mcp

CI pyright: strict Python 3.14+ License: MIT

MCP server for managing job application materials. Reads your resume from the workspace, fetches job postings and company info, compiles LaTeX resumes, and persists research, strategies, and interview prep per company. The host Claude does all reasoning — this server handles data access and persistence.

Quickstart

Requires Python 3.14+ and uv.

git clone <repo-url>
cd jobkit-mcp
uv sync

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "people": {
      "command": "uv",
      "args": ["run", "--directory", "/path/to/jobkit-mcp", "jobkit-mcp"],
      "env": {
        "JOBKIT_WORKSPACE": "/Users/you/Projects/resume"
      }
    }
  }
}

Related MCP server: jobjourney-claude-plugin

Tools

Tool

Description

get_profile

Read resume.tex from workspace root as professional profile

list_applications

List all application directories with their files

get_application

Get all text file content for one company

read_application_file

Read any file from an application by filename

save_application_file

Save/update any file in an application

delete_application

Remove an entire application directory

fetch_url

Fetch a URL, return text with HTML stripped

save_job_posting

Fetch URL or accept pasted text, save as job posting

compile_resume

Compile resume.tex with lualatex (company dir or root)

mock_interview

Start a mock interview — returns briefing for Claude to play interviewer

Workspace Layout

$JOBKIT_WORKSPACE/
├── resume.tex              # Base resume (used by get_profile)
├── coverletter.txt         # Base cover letter
├── META.md                 # Base strategy
├── stripe/
│   ├── resume.tex          # Tailored resume
│   ├── coverletter.txt     # Tailored cover letter
│   ├── META.md             # Application strategy
│   └── job_posting.md      # Saved job posting
└── anthropic/
    └── ...

Any file can be stored in a company directory — there is no whitelist. Build artifacts (.aux, .log, .out, .pdf, .gz, .fls, .fdb_latexmk) are skipped in listings.

Documentation

  • Tutorial — Walk through a complete application workflow end-to-end

  • How-To Guides — Focused recipes for specific tasks

  • Reference — Complete tool parameters, return values, and internals

  • Explanation — Architecture decisions and design rationale

Available Tools

10 tools
compile_resumeA

Compile resume.tex using lualatex. Compiles in a company directory, or the workspace root if no company is specified.

Args: company: Company name (optional — omit to compile the base resume)

Inputs: company, workspace_dir. Returns: result

delete_applicationC

Remove an entire application directory and all its files.

Args: company: Company name

Inputs: company (required), workspace_dir. Returns: result

fetch_urlB

Fetch a URL and return its text content (HTML tags stripped).

Args: url: Full URL to fetch

Inputs: url (required). Returns: result

get_applicationC

Get all text file content for one application.

Args: company: Company name (e.g. "stripe", "Anthropic")

Inputs: company (required), workspace_dir. Returns: files, company

get_profileC

Return the professional profile by reading resume.tex from the workspace root.

Inputs: workspace_dir. Returns: result

list_applicationsA

List all job application directories with their files.

Each application directory may contain resume.tex, coverletter.txt, META.md, job_posting.md, and any other research or strategy documents.

Inputs: workspace_dir. Returns: result

mock_interviewA

Start a mock interview session. Returns a briefing with all application materials, your profile, and interviewer guidance so Claude can play the interviewer role.

After the session, save feedback and areas to improve using save_application_file.

Args: company: Company name interview_type: One of: behavioral, technical, culture, manager

Inputs: company (required), workspace_dir, interview_type. Returns: company, profile, materials, interview_type, +2 more

read_application_fileB

Read a single file from an application directory.

Args: company: Company name filename: Name of the file to read (e.g. "resume.tex", "META.md", "job_posting.md")

Inputs: company (required), filename (required), workspace_dir. Returns: result

save_application_fileB

Save or update a file in an application directory.

Args: company: Company name filename: Name of the file to save (e.g. "notes.md", "company_research.md") content: Content to save

Inputs: company (required), content (required), filename (required), workspace_dir. Returns: result

save_job_postingB

Fetch a job posting URL or accept pasted text, and save it for a company.

Args: company: Company name url: URL of the job posting to fetch (provide url or content, not both) content: Pasted job posting text (provide url or content, not both)

Inputs: company (required), url, content, workspace_dir. Returns: result

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.3/5.0

    Scored across 10 tools

    Disambiguation4/5

    Tools have distinct purposes, but fetch_url and save_job_posting both involve fetching URLs; descriptions clarify difference (generic vs. job posting and save).

    Naming Consistency5/5

    All tool names follow verb_noun pattern in snake_case, e.g., compile_resume, delete_application, get_profile, mock_interview.

    Tool Count5/5

    10 tools provide a well-scoped set for job application management without being too few or too many.

    Completeness4/5

    Covers CRUD for applications, resume compilation, job posting saving, and mock interviews, but missing an explicit create_application tool to initialize a directory.

    Related MCP Connectors

    Related MCP Servers