career-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| CAREER_MCP_API_KEY | No | Groq or Gemini API key used by the bundled standalone CLI. Not required when using the MCP server with an MCP-capable client. | |
| CAREER_MCP_DATA_DIR | No | Override the location where resume, stories, and applications are stored. Defaults to ~/.career-mcp. | ~/.career-mcp |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| save_resumeA | Store the user's base resume, either as raw text or by reading a local file. Overwrites any previously stored resume. Provide exactly one of resume_text or resume_path. |
| get_resumeB | Retrieve the stored base resume. |
| save_voice_profileB | Store a description of the user's writing voice/tone (e.g. tone attributes, phrases to avoid, sample posts). Used to keep generated cover letters and LinkedIn About sections sounding like the user, not generic AI copy. |
| get_voice_profileB | Retrieve the stored voice profile, if one has been saved. |
| save_jdB | Store a job description for a company/role and mark it as the active application. Creates a new application record. Provide exactly one of jd_text or jd_path. |
| get_active_applicationA | Returns the stored resume, voice profile, and the active job description together, as context for match analysis or content generation. Use this to: (1) run match analysis — identify genuine overlaps and honest gaps between the resume and JD, not keyword stuffing; (2) draft a tailored resume by reordering/reframing existing resume content for relevance, never fabricating experience; (3) draft a cover letter; (4) draft a LinkedIn About section. Match the voice profile's tone if one is set. Save any generated output with save_output. Write the tailored resume as markdown in this exact shape so it renders cleanly to PDF: Full NameOne-line tagline (target title / positioning) Contact line: City · email · phone · linkedin.com/in/... (one paragraph) Summary (then a short paragraph)ExperienceJob Title — CompanyStart – End | Location
Skills (bullets, or a two-column table: | Area | Skills |)EducationOutput only the resume itself — no commentary before or after it. |
| list_applicationsA | List all stored applications (company, role, created date). |
| save_outputA | Persist generated content (match analysis, tailored resume, cover letter, or LinkedIn About draft) for an application. Defaults to the active application if application_slug is omitted. After saving a tailored resume, offer to export it with export_pdf. |
| export_pdfA | Render a saved markdown document to a PDF that ATS systems can read, using the user's installed Chrome/Edge/Chromium/Brave. Defaults to the active application's tailored resume. Only export a cover letter when the user asks for one. If no browser is found, an HTML file is written instead and the reason is returned — relay it to the user. |
| save_storyA | Save a career story directly (title, tags, content) without running the interview flow. Prefer start_story_interview for a guided capture. |
| query_storiesA | Retrieve saved career stories, optionally filtered by tag (e.g. leadership, conflict, failure-recovery, technical-deep-dive). Use to find a story relevant to a JD requirement, an interview-prep question, or a LinkedIn post idea. |
| start_story_interviewA | Begin a structured interview to capture one new career story. Returns the first question and a session_id. Keep calling answer_story_question with that session_id, relaying each returned question to the user verbatim, until status is 'complete' — the story is then auto-saved. |
| answer_story_questionA | Submit the user's answer to the current interview question and get the next one (or the completed, saved story). |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 13 tools
Most tools have clearly distinct purposes: pair-wise get/save for resume and voice profile, a separate story-interview flow (start/answer/query/save), and an output pipeline (save_output/export_pdf). The only real overlap is get_active_application, which re-returns the resume and voice profile already exposed by get_resume/get_voice_profile, so an agent must infer that the aggregate is for generation/analysis context rather than plain retrieval.
Every name follows a consistent snake_case verb_noun pattern (get_resume, save_jd, list_applications, query_stories, export_pdf), with no camelCase or style mixing. The story-interview trio (start_story_interview, answer_story_question, save_story) reads as a coherent family.
Thirteen tools is well within a comfortable range for a multi-stage workflow (resume, voice, JD, stories, generation, export). Each tool maps to a distinct step and none look like filler.
The write/read pairs and generation+export pipeline cover the main path, but the application lifecycle is thin: there is no update/get/delete for an application record (no status tracking, no switching active application other than by saving a new JD), and no way to list or retrieve previously saved outputs beyond export_pdf for the tailored resume. Agents will hit dead ends when managing multiple applications or reviewing saved drafts.