Job Search MCP Server

Job Search MCP

Your job search, with your AI assistant.

Find overlooked remote jobs, evaluate your fit, tailor your CV and cover letter, prepare for interviews and track every application. Connect your assistant to Hidden Jobs and keep your profile, CV versions, email alerts and application history together in your account.

Connect · Sources · Skills · Tools · Resources · Documentation

How it works

Your profile + selected CV
            ↓
 Find jobs in your sources
            ↓
      Compare + choose
            ↓
 Tailor CV + cover letter
            ↓
    Review + export PDF
            ↓
   Send your application
            ↓
   Track + prep interviews

Your assistant does the analysis and writing. Hidden Jobs supplies the offers and saves the profile, documents and progress you ask to keep. Use the same saved information from the connected clients you authorize.

Quick start

1. Connect your assistant

Add a custom remote MCP server in your assistant, enter the URL, sign in to Hidden Jobs and approve the connection.

Claude Code

claude mcp add --transport http hidden-jobs https://hiddenjobs.dev/mcp
claude mcp login hidden-jobs

Codex CLI

codex mcp add hidden-jobs --url https://hiddenjobs.dev/mcp
codex mcp login hidden-jobs

Complete sign-in in your browser, then return to your client. See client configuration for graphical clients, terminal verification and troubleshooting.

2. Set up your profile and CVs

Once connected, ask your assistant to build your profile and save your CVs.

Set up my profile from this CV. Save it as Backend English and ask me about any missing details.

Keep multiple CVs, such as Backend English and Manager Spanish, with immutable version history. Choose the CV and exact version for each application, or ask your assistant to choose the best match for the offer. Transfer files through your client's supported file capability, or upload them in Profile.

3. Find and compare jobs

Search Hidden Jobs and Remote OK for TypeScript roles open to applicants in Spain. Compare three offers against Backend English, version 3, and explain the best match.

Choose sources in search_jobs: Hidden Jobs, Jobicy, We Work Remotely, Remote OK or Himalayas. sources is required, with at least one selection and no default. Reading an offer or opening its link also requires the returned source. Use open_application_link to retrieve original application links. External offers are queried through public APIs/feeds and are not stored in Hidden Jobs. You can compare them and prepare documents; saved job evaluations, activity, application tracking and email alerts use Hidden Jobs offers. See job sources.

4. Prepare your application

For this Hidden Jobs role, use Backend English, version 3, to tailor my CV and write a cover letter. Review both, save the application and export the final CV as a PDF.

Compare all my saved CVs against this offer, choose the best match, explain why, and use it to tailor my CV and prepare the application.

Your assistant compares the content of your saved CVs against the role and uses the exact selected version. You can name a specific CV instead, and your default CV stays unchanged.

The assistant saves a snapshot of the offer and the chosen CV version, drafts documents from confirmed facts, reviews them and links the final documents to the application. Creating version 4 later does not replace the version already selected for that application.

You review and submit on the employer's website. Then ask your assistant to record the submission.

5. Track and prepare for interviews

I submitted this application today. Mark it as applied and save the next step.

Prepare me for the first interview using the role, company information and the CV I used.

Open Applications to see your pipeline, filter statuses and review history, documents, notes and next steps. Use Profile for your experience, CV library and connected applications.

Email job alerts

Create a daily alert for Backend or DevOps jobs in Europe matching TypeScript.

Change that alert to weekly. Pause it for now, then show me my other alerts.

Manage the same alerts as the dashboard: up to five saved searches, daily or weekly emails, and multiple filter selections. Creating, modifying or resuming requires a confirmed account email. You can also pause all alerts or delete them. See job alerts for filters, version conflicts and examples.

Skills

15 portable skills guide the complete workflow. Each directory contains a SKILL.md with instructions your assistant can follow.

For example: “Use apply to prepare this application, then review to check it.”

Install the selected skill directories using your client's skill support. A remote MCP connection does not automatically install local skill files. You can also request the same instructions as MCP prompts, workflow resources or through get_workflow; this supports clients with different interfaces.

See the skills guide for examples, the tools each workflow uses and how to run it from a remote chat.

Tools

45 tools expose job search, email alerts, private career data and document operations. Tools read or save information; your assistant evaluates offers, personalizes your CV, writes letters and leads interview practice.

For example, preparing an application uses get_cv → save_application → save_document → update_application. PDF export adds render_document, check_document and get_document_download_url.

See the complete tool reference for arguments, examples, permissions and errors.

Resources

Resources give your assistant reusable instructions, templates and access to your saved career information.

For example, hiddenjobs://workflows/interview explains how to prepare for an interview; hiddenjobs://me/applications/{id} supplies the saved application to work from. Private resources enforce the same OAuth and account-ownership checks as the tools. Use resources/templates/list to discover private URI patterns and the list/get tools to obtain record IDs.

The 15 MCP prompts use the workflow names: alerts, setup, expand, search, match, apply, review, export, interview, upskill, outcome, email-sync, report, add-template, reset. See the resources and prompts reference.

Your data and documents

• Profile facts, CVs, immutable CV versions, drafts, evaluations, interview notes, learning plans and application history are stored privately in your Hidden Jobs account, rather than relying on a local project folder.
• Your OAuth-authorized applications can work with your saved career data. Export your data or confirm deletion from your account.
• PDF, DOCX and UTF-8 text/Markdown uploads support up to 2 MiB per file and 20 PDF pages. PDF export supports Latin text and validated layout settings. Review the final PDF before sending it.
• Application states are saved, preparing, prepared, applied, interview, offer, rejected and withdrawn. Prepared documents are not sent automatically; submission and outcome changes are recorded with your confirmation.
• Email updates use your separately authorized email connector and your approval. Hidden Jobs keeps approved references and summaries, not mailbox credentials or email bodies.

See the career workspace guide for storage, document transfer, application tracking and templates.

Documentation

This repository contains the MCP protocol, REST implementations of every tool, external-source adapters in src/sources/, skills, resources and client connection guides. The remote connector uses the same HTTP adapter published here. The hosted backend and account dashboards are maintained in the Hidden Jobs application repository.

• Client configuration
• REST adapter and tool routes
• Skills and workflow examples
• Complete tool reference
• Resources and prompts
• Career workspace
• Protocol
• Architecture
• Development
• Security · Contributing

License

MIT. See LICENSE.

Employer names and identifying job slugs are withheld from search and detail results. get_company provides anonymous company context using a job ID. If an operation requires account action, the assistant receives a link to the account settings.